r-390-eng-txï¿½ · Title: r-390-eng-txï¿½ Author: Unknown Created Date: Saturday, May 01,...

SECURITY INFORMATION - RESTRICTED

RESTRICTED

Introduction

This project was initiated to fill a gap in the documentation that was electronically available forthe R-390A receiver.

In March, 1999, Bill Hawkins scanned the document into an ASCII character-based file and re-produced the content. I scanned in and added the readable sketches from among the 74 figuresin the document. Many of the figures were unusable because the originals were photographswith very poor registration. The document was then formatted using Windows “Courier” font(bold) in MS Word and transferred to an Adobe Acrobat Portable Document Format (“.pdf”).

Bill, as others have, has left in all of the original errors as a tribute to an unknown typist at theCollins Radio Company .

Known errors include:• , 4th paragraph -“these equipments” is as written. This style is used throughout

the document.• Page 7, 5th paragraph - “layed out” is as written• Page 8, 1st paragraph - “layed out” is as written• Page 14, 3rd paragraph - ‘hazzard’ is as written• Page 15, first line - ‘apprximately’ is as written• Page 18, 3rd paragraph - “minumum” is as written• Page 21, 2nd paragraph - “accommodate” is as written• Page 30, 1st Paragraph - “pehnolic” is as written• Page 35, 4th paragraph - “curcuit” is as written• Page 36, 1st paragraph - “Corp” is as written• Page 37, 4th paragraph - “lenthly” is as written• Page 42, 8th paragraph - “aggrevated” is as written• Page 47 3rd paragraph - “montstrocity” is as written• Page 52, last paragraph - “abondoned” is as written• Page 61, 9th paragraph - “consistant” is as written• Page 69, 3rd paragraph - “mocrophonics” is as written• Page 74, 3rd paragraph - “mocrovolts” and “coupled into the ground” are as written• Page 80, 1st paragraph - “extention” is as written

If there are other errors that need to be pointed out, or if I have introduced new errors, I wouldlike to know about then so that I can document them or correct them.

The original security classifications and warnings have been retained, even though thedocument was declassified in the mid-1960’s.

1999-April-18Al TirevoldWAØ[email protected]

1999-May-01 - Minor text corrections made


RESTRICTED

CER No. 297

FINAL ENGINEERING REPORT ON

RADIO RECEIVERS R-389( )/URR AND R-390( )/URR

SEPTEMBER 15, 1953

Object: Development of Low and Medium Frequency Radio Receiver

Signal Corps Cpntract No. W36-039-sc-44552

Signal Corps Specification - SCL-1134-B

Department of the Army Project No. 3-24-01-051

Signal Corps Project No. 15-805G-2

Placed by U.S. Army, Signal Corps Engineering Laboratories, Fort Monmouth, New Jersey

This document contains information affecting the national defense of theUnited States within the meaning of the Espionage Laws, Title 18, U.S.C.,Sections 793 And 794. The transmission or the revelation of its contentsin any manner to an unauthorized person is prohibited by law.

PREPARED BY: APPROVED BY:

/s/ L. W. Couillard /s/ [unreadable]____________________ ___________________Project Engineer Director of Engineering and Research Division

A PUBLICATION OFTHE RESEARCH AND DEVELOPMENT LABORATORIES

COLLINS RADIO COMPANYCedar Rapids, Iowa


RESTRICTED

TABLE OF CONTENTS

Sections Title Page

1.0 Purpose . . . . . . . . . . . 1

2.0 Abstract . . . . . . . . . . 1

3.0 Publications . . . . . . . . 4

4.0 Factual Data . . . . . . . . 4

5.0 Conclusions . . . . . . . . . 78

6.0 Recommendations . . . . . . . 79

7.0 Identification of Personnel . 80


RESTRICTED

LIST OF ILLUSTRATIONS

Figure Title Page

1 Typical Frequency Systems 2 Radio Receiver R-389/URR Final Mixing Diagram 3 Radio Receiver R-390/URR Block Diagram of R-F Unit 4 Radio Receiver R-389/URR Front View 5 Radio Receiver R-390/VRR Front View 6 Radio Receiver R-389/URR Rear View 7 Radio Receiver R-389/URR Top View - Shield Removed 8 Radio Receiver R-389/URR Bottom View 9 Radio Receiver R-389/URR Front View - Panel Removed 10 Radio Receiver R-389/URR Front View - Unit Removed 11 Radio Receiver R-389/URR Bottom View - Unit Removed 12 Radio Receiver R-389/URR R-F Unit - Rear View 13 Radio Receiver R-389/URR R-F Unit - Bottom View 14 Radio Receiver R-389/URR Gear Train 15 Radio Receiver R-389/URR R-F Coil 16 Radio Receiver R-390/URR Rear View 17 Radio Receiver R-390/URR Top View 18 Radio Receiver R-390/URR Top View - Shield Removed 19 Radio Receiver R-390/URR Bottom View 20 Radio Receiver R-390/URR Front Panel Removed 21 Radio Receiver R-390/URR Panel and Units Removed 22 Radio Receiver R-390/URR Bottom View - Units Removed 23 A-C Power Supply - Top View 24 A-C Power Supply - Bottom View 25 28 Volt D-C Power Unit 26 Audio Unit - Top View 27 Audio Unit - Bottom View 28 I-F Unit - Top View 29 I-F Unit - Bottom View 30 B.F.O. Unit 31 I-F Coil 32 Crystal Filter 33 Antenna Relay Unit 34 Antenna Relay - Cover Removed 35 Radio Receiver R-390/URR R-F Unit - Oblique View 36 Radio Receiver R-390/URR R-F Unit - Rear View 36A Radio Receiver R-390/URR R-F Unit - Bottom View

ii


RESTRICTED

LIST OF ILLUSTRATIONS (Cont.)

Figure Title Page

37 Radio Receiver R-390/URR R-F Chassis 38 Radio Receiver R-390/URR R-F Gear Train 39 Radio Receiver R-390/URR Antenna Coil 40 Radio Receiver R-390/URR R-F Coil 41 Crystal Oscillator - Top View 42 Crystal Oscillator - Oven Removed 43 Calibrator Unit - Top View 44 Calibrator Unit - Bottom View 45 Radio Receiver R-389/URR VFO 46 Radio Receiver R-389/URR VFO - Cover Removed 47 Radio Receiver R-390/URR VFO 48 Radio Receiver R-390/URR VFO - Cover Removed 49 Table Cabinet 50 Table Cabinet - Collapsed 51 A-C power Diagram 52 Dynamotor Filter Circuit 53 115 Volt D-C Inverter 54 115 Volt D-C Rotary Supply 55 Line Audio Circuits 56 Audio Response Curves 57 I-F Unit Block Diagram 58 I-F Transformers 59 Crystal Filter Circuit 60 Interstage I-F Transformer 61 Radio Receiver R-389/URR Trial Mixing Diagrams 62 Radio Receiver R-389/URR Trial Mixing Diagrams 63 Radio Receiver R-390/URR Antenna Input Circuit 64 Second Crystal Oscillator 65 Crystal Oscillator Coil 66 AGC Circuit 67 Carrier Meter Circuit 68 Radio Receiver R-390/URR VFO Temperature Curve 69 Noise Limiter Circuit 70 Radio Receiver R-390/URR VFO Transformer 71 Radio Receiver R-389/URR Mechanical System 72 Radio Receiver R-390/URR Mechanical System 73 Radio Receiver R-389/URR Schematic 74 Radio Receiver R-390/URR Schematic

iii


RESTRICTED

1.0 PURPOSE

This report covers all of the work done on the Signal Corps Develop-ment Contract W36-O39-sc-44552 which resulted in the development of theRadio Receivers R-389( )/URR and R-390( )/URR as specified in SCL-1134-B.The Signal Corps has plans for the development of a series of radio receiverswhich would provide reception over the frequency range of 15 kilocycles to3000 megacycles. The receivers covered by the specification SCL-1134-B aretwo of this series and cover the frequency ranges 15 kc to 1500 kc and 500 kcto 32 mc. The purpose of this project was to develop and build three modelsof each of these receivers along with their required accessories.

These receivers would fulfill many needs of the Armed Forces but ingeneral there were two basic requirements. First, to provide the increasedstability and calibration which is becoming almost a necessity in moderncommunication systems; and second, to combine into one general purpose re-ceiver as many as possible of the facilities and special circuitry that arenow being used by the various branches of the Army and which now requiremany different receivers to provide. Along with these receivers were de-veloped accessories such as special power supplies and cabinets. There werealso instruction books prepared and a complete set of manufacturing drawings.Throughout this program a great deal of effort was made to use the very bestand latest types of components and materials. Also very close contact wasmaintained with the Signal Corps Engineering Laboratories to fit these re-ceivers into their very latest plans for communication systems.

The purpose and responsibilities of this development project weredoubly emphasized when, after the first two engineering models were sub-mitted, a contract for production quantities of both R-389 and R-390 re-ceivers was received by Collins Radio Company. In addition, a secondcompany, Motorola, was set up as a directed subcontractor for a largequantity of R-390 receivers. These production contracts resulted in seriousresponsibilities being placed on the development work, since both labor andfactory space commitments were made by both companies. During the latterpart of the development program, a great deal of time and effort was divertedfrom development to production, thus aiding production but delaying thefinal phase of the development program.

Therefore, during the latter phases of this contract, the objective wasnot only to develop these equipments as specified, but also to aid and ensurethe delivery of production quantities.

2.0 ABSTRACT

This section contains a brief resume of the report to follow empha-sizing the difficulties encountered and the progress made.

The report covers the work done in fulfillment of the developmentcontract for the R-389 and R-390 receivers. Although it is divided into

1


RESTRICTED

seven sections, the main body information is contained in the section onFactual Data. The Factual Data section is divided into three main taskscovering the Planning, Models, and Final Report.

The Planning Section covers the work done before any actual modelbuilding could be started. This was actually a very important stage andtook considerable time, mostly spent on the RF sections, The main problemshere were spurious responses due to the complicated mixing scheme requiredfor stability. The final performance of the equipment did not meet thedesired 80 db level, but was still better than previous equipment andsatisfactory to the Signal Corps.

The Model Building Section contains a detailed description of allthe development work done on this contract and is divided into and coversvarious topics much the same as the equipment is divided into sub-units.These topics are as follows:

1. Mechanical Design

This covers the basic structural layout and mechanical drivesystems. The drive systems for both receivers were the most difficult ofthe mechanical design problems. A great deal of time was spent in workingout gear trains and mechanical drives which were not only easy to maintainbut also as rugged as possible. The systems finally used are much morecomplicated than similar receiving equipment but were the best meanspossible to accomplish the tuning required.

2. Power Circuits

This covers the design factors and performance of all three dif-ferent power units and their associated circuits. The electronic regu-lator design was the biggest problem due to the requirement for a plusor minus 15% input voltage, but was finally satisfactorily overcome. Theexternal 110 volt DC supply also required a lot of time since so manydifferent systems were tried. The rotary machine finally shipped on thecontract seemed to be the best compromise for this unit.

3. Audio Circuits

This covers the design of the audio circuits and their per-formance. The main problems were the filter and metering circuits whichwere overcome without too much difficulty.

4. IF Circuits

This covers the design of the IF circuits giving detailed designof various transformers. This section is rather long since the IF per-formance was considered as one of the major features of this equipment.The main problem in the IF was to get the desired bandwidths and to holdgood performance under temperature and humidity. The IF bandwidths turnedout quite satisfactorily, but the temperature and humidity problem wasnot completely resolved.

2


RESTRICTED

5. RF Sections

This covers the development of the R-389, R-390 RF sections in-cluding the R-390 crystal unit. The RF units were undoubtedly the mostdifficult to design of the entire equipment since they include bothmechanical and electrical problems. The electrical difficulties includedsensitivity, spurious responses and mixer design. All of these problemshave been worked out satisfactorily.

6. Variable Oscillator Designs

This covers the BFO and VFO design for both the R-389 and R-390.The main problem in all of these oscillators is stability which came outrather good although further improvement is being considered.

7. Crystal Calibrator

This covers the design of the 100 kc crystal calibrator unit inwhich the main problem was the multi-vibrator performance. This is nowconsidered satisfactory except for the possibility of circuit aging whichwill be investigated during production.

8. Special Circuits

This covers the work done on special circuits such as the AGCsystem, noise limiter and others. No major problems were involved inthese special circuits.

To complete this work it has taken approximately 30,000 engineeringman hours implemented with laboratory technicians and drafting help.This work has been spread over a three-year period with the deliveryschedule as follows:

Project Started June, 1949

R-390 No. 1 January, 1951

R-389 No. 1 March, 1951

R-390 No. 2 September, 1951

R-389 No. 2 February, l952

R-390 No. 3 February, 1952

R-389 No. 3 May, 1952

Miscellaneous Accessories February, 1953

Final Report August, 1953

3


RESTRICTED

The accessories and later items required by this contract which in-clude drawings and final report, have been delayed considerably, due tothe efforts being made to meet production schedules for some of these sameequipments.

In general, this report has stressed only the new and importantpoints of this work and has left many of the lesser details to the monthlyreports and instruction books. For a more complete record of this program,the other publications should be consulted.

3.0 PUBLICATIONS

The following is a list of publications which are directly concernedwith this project and may be used to supplement this report:

1. Engineering Proposal for R-219 Receiver - Collins Radio Co.

2. Monthly Reports - Contract DA36-039-sc-44352 - Collins Radio Co.

3. Instruction Book, R-389 - Collins Radio Co.

4. Instruction Book, R-390 - Collins Radio Co.

5. Manufacturing Drawings and Specifications - Collins Radio Co.

6. R-389 - R-390 Specification SCL-1134-B (Development)

7. R-389 - R-390 Specification MIL-R-10474 (Production)

8. R-389 Type Test Procedure - Collins Radio Co.

9. R-390 Type Test Procedure - Collins Radio Co.

10. R-389 Production Test Procedure - Collins Radio Co,

11. R-390 Production Test Procedure - Collins Radio Co.

12. 51J-3 Instruction Book - Collins Radio Co.

4.0 FACTUAL DATA This section contains a discussion of the work accomplished duringthis contract. This work can be broken up into separate tasks as follows:

Task A – Planning

Task B - Model Building and Testing

Task C - Drawings and Report

Most of the time was spent on task B which involved the building andtesting of six complete models with all their accessories.

4


RESTRICTED

4.1 Task A - Planning

The work of planning a project of this size is quite extensive andbegan with the first proposal submitted for bidding purposes. At thattime a detailed analysis of the specification was not practical and onlygeneralized planning was done. However, it was apparent at this time,that the basic requirement for these receivers, was a high degree ofstability and calibration. This stability and calibration was obtainableonly by special superheterodyne systems and the first major problem to besolved was the determination of the frequency schemes to be used.

In the engineering proposal for this development contract submittedin March, 1949, it was stated that the frequency scheme to be used inheterodyning the received signals down (or up) to a fixed IF frequencywould be similar in basic principal to the Collins, commercial, com-munications, receiver, the 51J. This basic principal is to build atunable, low frequency receiver covering a one megacycle range with ex-treme accuracy and stability. This low frequency receiver is then usedas a tunable IF strip for the higher frequencies by using fixed crystalsto heterodyne one megacycle ranges of the higher frequencies down to thetunable IF range. A typical heterodyne system, as described above isshown in Figure 1.

The above scheme is not usable throughout the entire frequency rangecovered by these receivers so special frequency schemes were needed belowabout 8 megacycles. The best way to cover these lower frequencies wasdetermined to be by use of "triple conversion." A typical frequencyscheme of this type is also shown in Figure 1.

Before any actual numbers for signal ranges and injection frequenciescould be decided upon, several points had to be carefully considered. Themost important of these were (1) spurious signals, (2) circuit compli-cations, and (3) mechanical problems. The specification called for allspurious to be down at least 80 db and this was the design goal. Therewas no actual limits on circuit complication although it was known thatreliability and ease of maintenance were important considerations. Themechanical problems for switching and tuning such a system could be verygreat, so mechanical simplification was given considerable thought duringthis planning phase.

The spurious response were first considered to be of two generaltypes, but later a third type caused serious trouble and some changeshad to be made. The first type occurs when the IF is higher than thesignal and is such that harmonics of the signal feed directly into theIF. This can be expressed simply that n Fs = IF where Fs is the signalfrequency and n is an integer and is considered as the order of thespurious. One of these occurring in Figure 4 would be at 3.0 mc in the3-4 band when the 5th harmonic of 3 mc would fall in the high IF (15 mc).This would be a 5th order spurious.

5


RESTRICTED

The second type of spurious occurs when the signal or its harmonicsbeat with the injection or its harmonic to give the IF frequency. Thiscan be expressed by n Fs ±± m Fx = ±±IF when Fx is the injection fre-quency and m is an integer. The order of this type of spurious is(n + m). This type occurs in Figure k in the second mixer at 15 mc when(3 x 15) - (4 x 12) = -3.0. This would be a 7th order (3 + 4) spuriousand would occur on every band.

During the early stages every attempt was made to devise frequencyschemes such that no spurious below the 7th order would present. How-ever, when mechanical problems and space limitations were considered,this limit was dropped to 5th order.

At this time it was also found that there was no standard methodof spurious response measurement. The following method was used through-out this program:

1. Set a standard audio output for a 5 microvolt input on signal frequency.

2. Tune generator to spurious frequency and increase to same audio output

3. Spurious response ratio is the ratio between spurious input and 5 microvolts.

When it was found necessary to allow 5th order spurious in thesystems, it was realized that the 80 db figure required in the speci-fication could not be met. However, it looked like a limit of 60 dbwould be possible and seemed to be acceptable to the Signal Corps asthe best possible compromise.

The third type of spurious response mentioned previously shouldactually be classified as "internal signals", since they require nosignal input. These are caused by the beating and mixing of the variousoscillators within the receivers and are sometimes called "tweets,""whistles," or "birdies." There was no exact specification on internalsignals and likewise no standard method of measurement. The method setup for measurement was to compare the output of the internal signal toan equivalent external signal and express its strength in microvolts.The original design limit set on internal signals was one microvoltmaximum, but considerable trouble was encountered in trying to meet thislimit. The latest models had approximately the following performanceon internal signals:

Total number 25

Less than 1 microvolt 20

1 to 2 microvolts 4

2 to 3 microvolts 1

6


RESTRICTED

The strongest of these signals occur when the receivers are tuned toan internal oscillator frequency or some oscillator harmonic. They alsomay occur when two oscillators or their harmonics differ by one of the IFfrequencies. Considerable planning and testing of various frequency schemeswas done before any of the actual models could be started.

Before much frequency scheme planning could be done, it was necessaryto determine the spurious response characteristics of mixers. This re-quired considerable time, but did show about what could be expected forvarious types of mixer tubes and various operating levels. The best com-promise found was to use a triode mixer which had a remote cut-off charac-teristic. This gave a low noise figure and fairly good spurious performance.The mixer tube chosen was a 6C4 which gave 60 db rejection to a fifth orderspurious. This was considered usable and so a fifth order was set as thelowest order cross-over to be allowed.

Several frequency schemes for the R-389 were actually built in bread-board style before they could be fully evaluated. A more detailed des-cription of these various systems is given under the R-389 RF section(4.2.5). The final system is shown in Figure 2.

The R-390 frequency scheme was broken up into two ranges. The high rangeabove 8 mc would be a conventional double conversion system similarto the commercial 51J receiver. Below 8 mc, triple conversion was to beused. A great deal of calculating was required for the low range to pickcrystals and high IF ranges that were usable. The main requirements wereto avoid low order crossovers and to have the high IF ranges for all sevenbands be within a 2:1 tuning range. The final R-390 frequency system islisted in Table I and block diagram is shown in Figure 3.

At the same time that the detailed planning of the RF portion wasbeing done, a thorough study of the specification was made and tentativeplans were layed out for all parts of both receivers. One of the firstproblems encountered was the physical and mechanical makeup required ofthese receivers. The specification called for an immersion-proof casedesign which presented many problems but mainly one of temperature riseand heat dissipation. Since no air could be circulated in or out of asealed case, the first thoughts were toward meeting this requirement byradiation of the case alone. Tests were made using a temporary case ap-proximately the correct size with an estimated power requirement of theequipment. These early tests showed that high temperature operation wasa very serious problem and would require special heat-exchange unite tokeep the receivers within practical temperature limits.

At this point the Signal Corps decided that the requirement of awater-tight sealed case should be dropped rather than add any large ex-pensive cooling system. This also fit in with their ideas for a separatesmaller receiver for mobile field use and limiting the use of the R-389and R-390 units to fixed station use. This mobile receiver was later de-veloped to be used as part of the GRC-19 equipment and was typed theR-392/URR.

7


RESTRICTED

TABLE I

R-390 Frequency Scheme

Freq. Coil 1st Mixer High 2nd MixerBand Range Range Crystal I.F. Xtal Inj. 0 .5 - 1 .5 - 1 9 9.5 - 10 12 12 1 1 - 2 1 - 2 8 9.0 - 10 12 12 2 2 - 3 2 - 4 10 12.0 - 13 15 15 3 3 - 4 2 - 4 12.6 15.6 - 16.6 6.2 18.6 4 4 - 5 4 - 8 7 11.0 - 12 14 14 5 5 - 6 4 - 8 8 12.0 - 13 8 16 6 6 - 7 4 - 8 9 13.0 - 14 9 18 7 7 - 8 4 - 8 10 14.0 - 15 10 20 8 8 - 9 8 - 16 11 11 9 9 - 10 8 - 16 12 12 10 10 - 11 8 - 16 13 13 11 11 - 12 8 - 16 14 14 12 12 - 13 8 - 16 15 15 13 13 - 14 8 - 16 8 16 14 14 - 15 8 - 16 8.5 17 15 15 - 16 8 - 16 9 18 16 16 - 17 16 - 32 9.5 19 17 17 - 18 16 - 32 10 20 18 18 - 19 16 - 32 10.5 21 19 19 - 20 16 - 32 11 22 20 20 - 21 16 - 32 11.5 23 21 21 - 22 16 - 32 12 24 22 22 - 23 16 - 32 12.5 25 23 23 - 24 16 - 32 13 26 24 24 - 25 16 - 32 9 27 25 25 - 26 16 - 32 14 28 26 26 - 27 16 - 32 9.66 29 27 27 - 28 16 - 32 10 30 28 28 - 29 16 - 32 10.33 31 29 29 - 30 16 - 32 10.66 32 30 30 - 31 16 - 32 11 33 31 31 - 32 16 - 32 11.33 34

With the temperature problem simplified, the planning of the entiremechanical layout could be started. Most of the basic circuits for theentire receiver had been layed out and the approximate space requirementswere figured. Since unitized construction was to be used, the first problemwas to combine the basic circuits into units, keeping in mind the desira-bility of interchangeable units between the R-389 and R-390 models. Threecommon units were arrived at and were the power, audio, and IF units.

8


RESTRICTED

The variable frequency oscillators were to be the same size as thestandard Collins VFO except for the addition of the temperature controlovens. This left the two RF units plus a few miscellaneous items to befitted into the allotted space.

At this point, the planning phase of the project was considered atan end, and the building of models could be started.

4.2 Task B - Model Building and Testing

The contract called for one experimental model of each receiver tobe delivered followed by two engineering models of each. The actualprogram started with a paper mock-up model used to locate sub-assembliesand allot space for each. This paper model was followed by "bread-board"models which were the first attempt to package the bread-board circuitsinto their proper sub-units. These models were never really functioningmodels and were never delivered to the Signal Corps. The next modelsbuilt were the so-called engineering models and were actually deliveredto the Signal Corps for their comments. Since there were many knowndeficiencies in these engineering models, the design of the next modelswere actually started before the engineering models were delivered. TheSignal Corps engineers were working very close with the contractor atthis time so many of their opinions and changes were incorporated intounits before actual submission. Rather than build the second and thirdmodels together it was decided to build only the second and hold thethird models for any changes which were required by number two. Thisundoubtedly delayed the program but resulted in the third models beingmuch more complete and more acceptable to the Signal Corps.

4.2.1 Mechanical Design

The mechanical design work on these receivers was not started until theelectrical planning was far enough along to give good information as tocomponents and their location. One of the first things investigatedwas the case design which was specified to be water-tight. This meantthe problem of cooling had to be considered since excessive temperaturerise could be very serious. A power dissipation of 100 to 150 watts wasestimated and temperature rise measurements were made using a temporarycase of the approximate construction. These tests showed that thetemperature rise was way too high unless a very effective cooling systemwas also used. Rather than complicate the equipment with these extraparts, the Signal Corps dropped the requirement for a water-tight casein exchange for both a collapsible table cabinet and a rugged vehicularcase. Both of these cases could be ventilated so the cooling problemwas simplified considerably.

The next step was to build a paper model of the equipment, and workout the space and positions for the various sub-unite. This step re-sulted in the use of an "H" type frame on which units mount from both topand bottom. This basic design can be seen in Figures 21 and 22. Alsomounting on this frame were the front and back panels, and the top andbottom dust covers.

9


RESTRICTED

Another requirement which influenced the design of this frame was thepossibility of adding automatic tuning to the R-390 at some future date.Space for the automatic tuning heads and motor was left in the equipment anprovision for a separate casting for mounting these parts was made. Theaddition of the motor and "autotune" heads was done later for the FRR-33equipment and this model of the receiver was called the R-391.

The actual construction of the frames is rather unique in that it ismade up of six formed pieces, spotwelded together to give its final shapeand giving double thickness at all points. The frames for the R-389 andR-390 are identical except for differences in unit mounting holes so thatthe same forming tools can be used.

Captive screws on the sub-units were used to hold the units to theframe. Flush mounting nuts which pressed into the frame were used formating with these captive screws. Number ten mounting screws were usedand gave very little trouble in shock and vibration tests. These mountingscrews can be clearly seen in several of the unit photographs such asFigure 27.

The units are connected together by a main cable which is a part ofthe front and rear panels. This allows the cable and panels to be builtseparately and then attached to the frame.

It should also be noted in Figures 4 and 5 that the handles on thefront allow the equipment to be rolled over or stand on its face withoutdamage to controls. This is also carried out on the rear panel, Figures5 and 16, where bumpers are placed on either aide to protect the rearterminals and allow the equipment to be set down on the back withoutdamage. These back bumpers also have holes which engage pins in the ve-hicular applications in order to hold the back end better. The rear viewof these receivers in Figures 5 and 16 also show the location of the toolsand tube pin straighteners which were included. The plate showing in thelower right rear of the R-390 covers the hole which is used for a remotecable when converted for automatic tuning. There were no provisions madefor remote tuning of the R-389, since no requirements for this type ofoperation could be forseen.

Mounting on the top and underneath the main frame are seven mainsub-units. These are (1) power units, (2) audio unit, (3) I-F unit,(4) RF unit, (5) crystal oscillator, (6) calibrator and (7) variablefrequency oscillator. These units ended up in their various positions dueto a fairly logical arrangement. The VFO which is the main frequency de-termining unit, required direct coupling to the main tuning knob and sowas placed directly behind the tuning knob. The obvious place for this tuningknob was in the center of the panel where it could be easily reachedfor manual tuning. Since there were mechanical linkages necessary to theRF tuning, the RF unit was placed just above the VFO and extend to theright side. This allowed the band change knob to connect to the RF unitand still be on the lower right of the panel for ease in operation. Thisalso allowed the frequency counter to be part of the RF unit and still be

10


RESTRICTED

in the center of the panel where it should be. The crystal oscillatorunit required band switching and so was placed directly behind the RF unitand coupled into the band change drive. The shaft was also brought outthe rear for synchronizing the oscillator and for power to band switchexternal units. This feature was later used for driving the URA-13 ex-citer unit and the three mounting holes for this attachment can be seenaround this shaft hole.

The only other mechanical linkage necessary to be brought out throughthe front panel was the IF bandwidth switch and the BFO pitch control. Sincethese were both on the IF unit, this was placed on top next to the RF.The other units which had no mechanical drives were placed underneath theframe as shown. One very desirable feature which came out of this arrange-ment was that all normal alignment, both RF and IF, can be made from the topwhile the crystal oscillator trimming condensers are available from theback.

The power unit for AC operation is shown in Figures 23 and 24. Thethree captive mounting screws can be seen along with the four tapped holesin the power transformer itself, which fastens the transformer directly tothe side of the main frame giving extra support to this heavy unit. Theguard bracket around the plug should also be noted. The DC power unit isshown in Figure 25. This unit mounts by two screws into the front side andtwo captive screws accessible through the holes seen in the top. The sameframe mounting holes are used for both AC and DC units. The dynamotorhas rubber isolation washers under its cradle to reduce noise and the boxed-in portion contains filtering components for the electrical leads.

The audio unit is shown in Figures 26 and 27. The mechanical con-struction of this unit is straight-forward except for the shelf in thechassis to hold the large regulator tubes. The sockets on this shelf areof a very flat design to allow for the extra tube height. The holes aroundthese big tubes are for ventilation.

The IF unit is shown in Figures 28 and 29. The sealed unit under thechassis is the BFO and is shown opened in Figure 30. This unit is turnedthrough the "bellows" type coupler since the shaft is threaded and movesin and out of the oscillator. The construction of some of the IF trans-formers are shown in Figures 31 and 32.

The R-F units were started as soon as their frequency schemes weredetermined. These units were by far the most complicated part of theequipment both electrically and mechanically. The basic mechanical problemwas to tune and switch the required number of circuits in the allottedspace. Tuning of the RF circuits by means of moving iron cores had beendecided upon previously, since this was the easiest method for handlingso many circuits and it also eliminated the design of a special tuning con-denser. To cover the necessary tuning range in the R-389 coils, approxi-mately 1-1/4 inches of motion was required. This ruled out cams forlifting the tuning rack since there was not enough room to swing cams thissize. To do the job of moving the racks, reversible lead screws similar

11


RESTRICTED

to those used in level wind fishing reels were used. These can be seen inFigure 12. Figure 13 shows the worm gears used to drive these screws suchthat different worm and gear ratios allow the various racks to be moved atdifferent speeds while all are being driven from the same tuning knob.Figure 71 is an exploded view of the a-389 mechanical system showing de-tailed operation.

In the center of the RF unit is a motor-driven band switch which auto-matically selects the proper coil range as the receiver is tuned. This motoris operated by a commutator whose various segments are arranged to corre-spond to the desired bands. This entire tuning and bandswitching system isdriven by the gear train shown in Figure l4 which also drives the frequencycounter or dial. In order to speed up the tuning of this receiver, whichwas rather slow due to the large number of moving parts, a tuning motor wasused and can be seen in Figure 13. This motor tuning allows the entirerange of the receiver (200 turns) to be covered in less than 40 seconds.It can be seen in Figures 12 and 14 that ball bearings are used in manyplaces to reduce the load on the tuning shaft. Considerable testing wasdone on the reversible lead screws and followers to find the best materialsand lubricants to reduce drag and wear. The final lead screws are chromeplate brass with hardened steel followers. The construction of the R-389RF coils is shown in Figure 15. The coil itself is covered by a largepowdered iron sleeve seen behind the trimmer coil. A tuning slug is in-cluded in this photograph to show how silver bands are placed around theslug to lower the circuit "Q" as it goes into the coil. The slug wireshown is a phosphor bronze spring which is necessary to reduce shock andprevent breakage.

The R-390 RF unit is shown in Figures 35, 36 and 36A. This unit alsouses permeability tuning, but only requires 0.800 inches travel. Thisallows the slug racks to be driven by means of cams. Both the cams and camfollowers can be seen in the rear view of Figure 36. The design of thesecams is such that 300 deg rotation is used for accurately lifting and po-sitioning the racks while the other 60 deg is used as a quick return. Thepositioning portion gives a linear rise while the return portion is de-signed for minimum torque requirements. These cams all operate in a pre-cise sequence and the marks for synchronizing them can be seen on the rearplate. Adjustment of each pair of came is possible since they are drivenby a gear which has a split hub and is clamped on its shaft.

The exploded view of the R-390 mechanical system is shown in Figure 72which shows the entire detailed operation. A rather special design is theGeneva and overtravel coupler used to drive the RF band switch. Thismechanism allows the 30 position megacycle knob to switch the six RF coilranges in and out at the proper time. The R-390 also uses a dial correctorwhich allows the drive to the frequency counter to be released through aclutch while the receiver is set to correct frequency. The frequency counteror dial is a part of the RF gear train and required special development bythe manufacturer of a geneva type transfer in order to meet the speedrequirements of the automatic tuning system. In order to reduce error and

12


RESTRICTED

backlash, most of the gears in this unit used are of the spilt-loaded type.There is one differential gear assembly used and in the case of the firstIF tuning, a differential motion is achieved by actually moving the coilcans slightly with one cam while the slugs are driven by another.

For ease in assembly of the R390 RF, the gear train can be builtseparately from the RF chassis as shown in Figures 37 and 38. The coilassemblies can be plugged in and typical assemblies are shown in figures39 and 40. The variable pitch of the winding necessary for straight linefrequency can be seen in these pictures.

During the vibration tests on the later models of the R-390, severalthings showed up which required changes in the future units. For one thing,the loaded gears caused considerable wear due to their "chewing" actionwhen driven by brass pinions. The pinions were changed to steel to relievethis problem. It was during these vibration tests that the need for thespring wires on tuning cores instead of a solid wire was found. There wasalso some indication of the rack followers "pounding" the brass cams causingdents, but this was not considered serious enough to warrant steel camssince the cutting of the softer brass cams is less of a manufacturingproblem.

When this RF unit is placed in the main frame, its gear train mesheswith the two knob gears which are on the lower casting, The tolerancesinvolved here are rather large and a procedure for "shimming the castingwhen it is mounted into place is required. The RF unit itself is locatedby its front gear plate resting on the bottom the the main frame and beingheld against the side of the frame by two side screws. A better way to tiethese two gears together would have been to use a common gear plate but therequirements of interchangeability with the automatic tuning system madethis impossible without considerable redesign.

The normal use for both equipments is in a fixed station where theywould be mounted on a relay rack in a normal fashion. However, specialcases were designed for use in table mounting and vehicular use. The tablecabinet is shown in Figure 49 and can be taken apart for shipment as shownin Figure 50. The vehicular case is very similar in design except formore rugged design which includes permanent welding instead of beingcollapsible.

This covers the general mechanical design of the equipment and thevarious circuits and subassemblies will be taken up separately.

4.2.2 Power Units

There were two power supplies developed for these receivers under thiscontract. One for 115/230 volt AC operation and one for 28 volt DC oper-ation. These two units are completely interchangeable except that powercords and line fuses must also be changed. They both fit into the samespace in the equipment and are mounted in the same holes.

13


RESTRICTED

4.2.2.1 AC power Units

The AC power supply unit is shown in Figures 23 and 24. A complete powerschematic is shown in Figure 51.

Early in the program it was decided that the most efficient way, inboth space and weight, to supply plate power for the receiver's operation,would be to use electronic regulation throughout. This was dictated byseveral things. First the need for close voltage control on all oscil-lators in order to meet the stability requirements. Second, the need forclose voltage control or, amplifiers stages since gain variations were tobe held to a minimum. Third, the space and weight required would be lessthan if condensers and chokes were used to do same amount of filtering.Fourth, the transformer voltage would be lower than if the same voltageregulating were done with V-R tubes. Fifth, the actual power dissipationwould be lower than with a comparable choke, condenser, and V-R tubesystem. One disadvantage of an electronic regulated system would be theextra tubes and circuits needed, however, this was not considered to bea serious objection.

The decision to use 180 volts for the plate supply voltage was in-fluenced by several factors. First, it was desirable to keep voltage aslow as possible because of heat dissipation, component size and rating,insulation problems, and shock hazzard. The factors requiring high voltagewere power output requirements, oscillator performance, and the front-endrequirements on blocking and cross-modulation. The value of 180 volts wasselected as a compromise between these various factors.

Early in the program a B+ current requirement of 125 ma was esti-mated and the first power supply circuits were built accordingly. Asthe equipment was developed more and more current was required until thelast models ended up using about 200 ma. The supply shown will deliver200 ma at 180 volts with excellent regulation and low ripple contentas the following data indicates:

Input voltage variation of ±±15% B+ variation ±±l volt maximum

Load current variation 100 to 200 ma B+ variation ±±2 volt maximum

Ripple voltage normal 0.01 volts

Ripple voltage worst condition 0.1 volts

In order to get a transformer to handle this amount of power in asmall size, Class B insulation was used in transformer. This allows thetransformer to run at 125 deg C at its hottest point.

Some of the early, lower current models used a 5U4G for a rectifiertube, however, the latest models use two 25Z5W tubes. In order to reducethe peak current and to balance the parallel sections, a 47 ohm resistor

14


RESTRICTED

is placed in series with each cathode. This supply feeds apprximately250 volts into the electronic regulator circuit as shown in Figure 51.Two 6082 tubes in parallel serve as series regulating elements and arecontrolled by a 6BH6 DC amplifier coupled to their grids. This 6BH6 isfed from the regulated B+ through two 5651 gas tubes which present aconstant voltage drop. Thus any variation from the 180 volt value isamplified by the 6BH6 which changes the 6082 bias so as to bring thevoltage back to 180 volts. AC filtering is accomplished first by thelarge input condenser on the regulator plates and also by feeding outof phase ripple back through the 6BH6 amplifier. This is done boththrough its grid circuit and plate circuit with a "hum balancing" po-tentiometer in the plate coupling circuit. This circuit worked very wellboth for AC filtering and also serving as a low impedance for AC in theB+ circuits. However, an extra 10 mf condenser was later found neces-sary on the B+ line to reduce audio coupling through the power supply.

The 6082 tubes were developed by RCA on a Signal Corps developmentcontract and were very desirable because of their low plate impedance,small size, and high current capacity. The 47 ohm resistors in eachcathode lead serve to balance the parallel sections. Some of the earlymodels were built using only one 6082 but were found to be inadequate forthe amount of power dissipation. Two tubes just handle the power whenthe voltage is maximum. The heat developed by these tubes is quite highand every possible means of cooling has been done. Even so all com-ponents in this area were selected for maximum temperature operation.These tubes mount on the audio unit upside down on the lower side ofthe equipment which also made the heat problem worse. Some trouble wasexperienced under vibration with the 6082 tubes due to foreign piecesof material inside causing momentary shorts. The manufacturer is in-vestigating this and hopes to find a remedy.

Also in the power unit is a 6 volt dc supply for the antenna, break-in, and squelch relays. This supply is a bridge type circuit using acopper-sulfide rectifier. The main problem here was effect of heat andmoisture on the rectifier itself. These problems were solved by de-signing the relays for a minimum current drain and by adding extrafinish coats of varnish to the rectifiers. Very little trouble hasbeen experienced with this circuit.

The filament regulation circuit for the oscillator filaments shouldalso be covered here although this ballast tube is mounted on the IFunit. Considerable work was done with Amperite Corp. in designing thisspecial ballast tube which feeds a constant 300 mils to the two 6 volt300 mil oscillator tubes used in the VFO and BFO. These three tubesare connected in series across the 26 volt filament supply. The ballasttube (3TF7) operates on a current of 290 to 330 mils and holds this cur-rent within ±±10 mils for input voltage variations of ±±15%. This reducedthe 15% variation to approximately 3%. There is some question if afilament regulator is necessary in these receivers since the oscillatorsare very good even without regulation. However, since the stability wasa big factor in this design and since the factor of tube aging was notknown, the regulator was included.

15


RESTRICTED

4.2.2.2 28 Volt D-C power Unit

The 28-volt DC power supply is shown in Figure 25. The completeschematic of this unit is shown in figure 52.

The design of the 28 volt DC power supply was a relatively simplematter except for the noise filtering problems. The requirement forthis DC operation was the reason for using 28 volt filament stringsthroughout the units. The design of the dynamotor itself was no problemother than changing the current rating as the receiver load increasedduring the program and in keeping the overall size down. The finalmodel of this dynamotor was supplied by the Electrolux Corp. of Con-necticut.

The problem of dynamotor brush noise, especially when used withthe low frequency receiver, was quite serious. Several different filterdesigns were tried before an adequate circuit was found and even thenit was felt that some improvement could be made by further work.

This present 28 volt DC supply does cause a slight drop in signal-to-noise ratio at certain frequencies in both receivers, about 2 db inthe R-390 and 3 db in the R-389. There is also some noise radiationback through the filament and power circuits, but with adequate shieldingbetween these leads the antenna, trouble can be eliminated.

This dc power unit is interchangeable with the AC unit and uses thesame electronic regulator system for the plate voltage. A special powercord and special fuses are required for dc operation. The power plug onthe back of the radios have four pins, two of which are used for ac andtwo for dc operation. The negative side of the DC is always connecteddirectly to ground. When operating from a 28 volt source, the R-389and R-390 draw about six amperes for normal operation. When the oscil-lator ovens are turned on an additional four amperes are used inter-mittently, depending upon the heat necessary. The life expectancy ofthis dynamotor is about 1000 hours.

4.2.2.3 115 Volt DC Operation

The equipment specifications also required a separate power unit for115 volt D-C operation. An investigation into the possibility of usingthe R-389 and R-390 receivers on a 115 v dc source indicated the powerunit would have to conform to the following requirements:

1. Input voltage: 115 v dc ±±15%. For this 15% change in input voltage, ac output voltage must not vary more than 15%.

2. Output power: 300 watts

16


RESTRICTED

3. Wave shape: Sinusoidal, with no RF noise or harmonics at frequencies from 15 kc to 32 mc in excess of 100 microvolts (peak volts). Radiated noise must be below 10 microvolts per meter at a distance of 2 feet (15 kc to 32 mc).

4. Frequency stability: Output frequency may not exceed the limits of 50 to 60 cps over the voltage and temperature range, and from no load to full load.

5. Ambient temperature: -40 deg C to +65 deg C operating, -62 deg C to +65 deg C non-operating.

6. Humidity: Operation after storage in humidity up to 100%.

7. Altitude: Transportation at altitude of 35,000 feet, with operation up to 10,000 feet.

8. Unit to be fungus-proofed per MIL-T-10513.

9. Expected life should be 1000 hours or greater without parts replacement.

The first type of inverter investigated was a vibrator unit, and acommercial unit was purchased for test purposes. Although the power out-put was adequate, the voltage regulation was not too good. The vibratorharmonic noise at 15 kc to 500 kc was excessive, and tests indicated thatit would be extremely difficult to filter this unit to prevent noiseradiation at these frequencies. This is due to the basically non-sinu-soidal wave shape obtained from the vibrator, which results in many har-monics of the vibrator frequency appearing in the output voltage. Vi-brators would have to be replaced about every 1000 hours. These vibratorshave a net price of about eight dollars.

The second type of inverter investigated was a thyratron supply. Thethyratron system has the advantage of no moving parts, and can be made tooperate over the required temperature range satisfactorily. Systems forregulating the output voltage and frequency could be devised which wouldbring them within specification limits. However, there are several dis-advantages to a thyratron inverter. The unit becomes fairly complex inorder to meet all of our requirements and the component parts are quiteexpensive and bulky. Thyratron tube life could not be guaranteed forlonger than 1000 hours and these tubes have a net price of about $25 each.Two tubes would be used per unit. RF noise measurements indicated thatit would be fairly difficult to get adequate filtering of the output. Aproduction quotation for 1000 of this type of unit was about $450 each.A schematic diagram of the Collins experimental inverter is shown inFigure 53.

The third type of inverter investigated was a small commercial rotaryinverter. The unit had an internal governor which regulated the outputvoltage to within 3% for a 15% input change. Output frequency is likewise

17


RESTRICTED

regulated by the governor to ±±3%. RF noise and harmonics were checkedusing an R-389 receiver, and it was found that with an antenna connectedto the receiver, the inverter had no effect on the noise level of thereceiver. The receiver was tuned over its complete range between 15 kcand 1500 kc., checking for noise and harmonics, and operation was con-sidered satisfactory. An R-390 receiver was also used and the noisechecked satisfactory from 500 kc to 32 kc. With proper lubrication,temperature extremes should have no adverse effect upon the inverter. Ithad been moisture and fungus proofed per MIL-T-10513. A production quo-tation of about $155 each for this type of unit in quantities of 1O00 wasobtained. The estimated life of this machine would be 5000 to 10,000 hours.Mechanical noise is somewhat of a problem in a rotary inverter although itcould probably be mounted in a location far enough from the receiver to beunobjectionable .

Of the three types of unite tested, it appears that the best compro-mise would be the rotary machine. It can be made to satisfactorily meetall of the requirements outlined for this equipment and is cheaper to buildand maintain than the thyratron supply. The vibrator supply is not ac-ceptable because of the high RF noise level it generates. Six of the rotaryconverters manufactured by Eicore Inc. of Chicago were shipped to the SignalCorp on this contract. An outline drawing of these units is shown inFigure 54.

4.2.3 Audio Circuits

General Design

The AF unit was designed to provide the required output within thespecified distortion limits with a minumum of power drain. Provision ismade for switching in suitable audio filters to give either a peaked fre-quency response at 800 cps or a frequency response which cuts off sharplyabove 3500 cps.

First Audio Stage

One half of a 12AU7 was chosen as the first audio amplifier. Oper-ation with an unbypassed cathode reduces distortion and allows a greatergrid input voltage swing. This is important because this tube is operateddirectly from the detector circuit without an intervening gain control. Itis so designed that the last IF tube will overload before this first audiostage begins to draw grid current. Since it was most convenient to usefilters in the medium and sharp positions having impedances of 600 ohms,the first audio stage is transformer coupled to these filters. An audioresponse selector switch selects the proper filter and/or attenuator tokeep the output constant (at 800 cps) irregardless of the switch position.The output of the filter in use is fed to two parallel 2500 ohm audio gaincontrols. One of these gain controls is shunted with a 1200 ohm resistorto present a combined load of 600 ohms to the audio filter.

18


RESTRICTED

Local Output

One half of a 12AT7 grid is fed from the local gain control. Itsplate is coupled to the grid of the 6AK6 local output tube. Negativefeedback is employed from the plate of the 6AK6 to the cathode of the12AT7 to reduce the output impedance. This negative feedback reducesthe voltage gain considerably and in order to bring it back up to areasonable level a small amount of positive feedback is provided bymaking a fraction of the cathode resistors of the half 12AT7 and 6AK6common to both tubes. The effect of the positive feedback is to can-cel the negative current feedback developed across the cathode of thehalf 12AT7 and so raise its gain.

Line Output

The other half of the 12AT7 has its grid fed from the line gaincontrol. Its plate is coupled to the grid of the 6AK6 line output tube.In order to keep the input voltage about the same for rated output onboth line and local channels, the plate resistor for the line output halfof the 12AT7 was reduced to 82K ohms. This requires a slight juggling of bothpositive and negative feedback resistors to obtain the same condition ofoutput impedance as for the local output.

Originally, the line output tube was a 6AK6 triode connected, butSignal Corps requirements for a -10 db position of the Line Meter switchnecessitated 10 db more available power which made it necessary to go topentode connections for this tube. Connected between the line outputtransformer and the line is a 14 db pad. When 10 mw is delivered to theload, the tube is actually putting out 230 miliiwatts. When the LineMeter switch is in the -10 db position, the line meter is connected di-rectly across the line output transformer secondary. Under this con-dition, the amount of power delivered to the line is 10 db less thanthat indicated by the VU meter. When the Line Meter switch is in the 0position, the VU meter reads the amount of power delivered to the line.Under this condition, the VU meter is connected to the output trans-former through a 10 db pad. When the Line Meter switch is in the +10position, the amount of power delivered to the line is 10 db more thanthat read by the meter. Under this condition a 20 db pad is insertedbetween the output transformer and the line meter. The electricaloperation of this switch is illustrated in Figure 55.

Squelch

The squelch threshold has been adjusted so that the squelch relayoperates at about 5 volts on the diode load. With the RF gain controlset at maximum, this corresponds to a signal strength of about 1 uv.The squelch relay is operated by the other half of the 12AU7, its gridbeing connected through an audio filter to the diode detector.

19


RESTRICTED

Output Tube Decoupling Filter

When a strong signal is tuned off to the side of the receiver se-lectivity curve, any random noise pulse is likely to set the receiverinto a type of audio oscillation which depends on the continued presenceof the external carrier for its maintenance. Audio is fed back from theoutput tubes to the B+ line which modulates this carrier. This is de-tected and fed to the audio and in this manner the oscillation maintainsitself. Placing an extra decoupling filter between the output tubes andB+ broke the feedback circuit and stopped the oscillation. However,temperature requirements of the receivers necessitated use of a paperfilter capacitor in the decoupling filter and it was impossible to getone large enough in the available space to completely eliminate cross-talk between the line and local channel.

Break-in Relay

A break-in relay is used to silence the audio unit when an associ-ated transmitter is in operation, and the receiver is being operatedunder "break-in" conditions. This completely eliminates any signalleakage which gets past the antenna break-in relay. Connecting thisrelay to the 600 ohm output from the audio transformer of the firstaudio stage eliminates hum pick-up troubles which have been experiencedwhen this relay is used in a high impedance grid circuit.

Audio Output Impedance Tolerance

The output impedance at the line output is 600 ohm ±±10% at alltimes. This is well maintained largely due to the presence of the pre-cision 600 ohm pads on the line output channel.

At the local output channel the output impedance varies over asomewhat wider range, due to variations in feedback resistors, tubes,etc. On this channel a figure of 700 ohms ±±20% is attainable.

Hum Level

A hum level of at least 50 db below the audio output is attainablefor any audio level above 1 mw. While this does not meet specificationsit is at least as good as broadcast transmitter hum levels which meansthat under ordinary use, the hum level at the audio output will be lessthan 3 db above the hum level sent by the transmitter.

Audio Output

The local output is at least 500 mw at less than 10% distortion at400 cps.

Although originally the line output distortion was also within specs(3% at to mw) but the necessity of placing a -10 db position on the linemeter switch results in a 10 db waste of power, which with the other 4 dbwaste required to make the VU meter reading correct, results in a dis-tortion at line output which reaches a maximum of 5% under some conditions.

20


RESTRICTED

Overall Audio Response

The original spec, SCL 1l34-B, called for a 2000 cps low pass audiofilter and a 1000 cps band pass audio filter. Before much work could bedone on these, the spec was changed and a 3500 cps low pass filter wassubstituted for the 2000 cps low pass filter and the center frequency ofof band pass filter was changed from 1000 cps to 800 cps.

The 3500 cps filter called for in the new specification, MIL R 10474(Sig. C) was to be equivalent to Federal #12 1008B. A filter was procuredand loaned to two filter suppliers. Both companies submitted satisfactorysamples. Later it became necessary to add another series regulator to theaudio chassis, and in order to accomodate it, the size of the filters hadto be reduced. The problem was presented to the suppliers and they bothcame up with a design which cut the size in half with only a slight dropin the response characteristic. The final response curves are shown inFigure 56.

Because of space limitations, the 800 cps band pass filter does notquite meet the frequency response curve specified but was approved by theSignal Corps. The case of this filter was also redesigned to be identicalwith that of the 3500 cps filter in order to conserve space on the audiochassis.

Photographs of the final audio unit are shown in Figures 26 and 27.

4.2.4 I-F Circuits

The design of the intermediate frequency amplifier for the R-389 andR-390 Radio Receivers was based on specifications included in the originalSignal Corps contract. These specifications included a complete de-scription of each selectivity position (six in all) by means of bandwidthat several attenuation points. It also included several other items suchas envelope phase delay, auxiliary circuits, and service condition.

4.2.4.1 Specifications and Limits

The original specification called for six selectivity positions, namedfor the bandwidth at 6 db attenuation points. These were 0.1, 1, 2, 4, 8,and 20 kc positions. The 20 kc position was later modified to 16 kc at3 db, and the 8 kc to be 8 kc at 3 db instead of 6 db. Bandwidths weregiven at 3, 6, 20, 40, 60, 80 db attenuation points for each selectivityposition.

The following table lists the various positions along with the nominalbandwidth and shape factor for each. Shape factor is defined as the ratioof 60 db BW to 6 db BW.

21


RESTRICTED

TABLE II

BW Position BW - 6 db Shape Factor

0.1 kc 0.1 kc 48

1.O l.0 6.8

2.0 2.0 4.25

4.0 4.0 2·55

8.0 11.5 1.91

16.0 17.0 1.82

The other important electrical specification was envelope phasedelay. This specified that the phase shift of a 150 cps modulating signalshould not vary more than 4.63 microseconds (1/4 deg) over the top band-pass (3 db pts.). This variation in phase shift of the modulating signaloccurs because of non-linearity in the phase-shift vs frequency curve forany one or group of tuned transformers. Specification originally calledfor this requirement in the 2 and 4 kc BW positions. This was later modi-fied to include only the 2 kc position. It was shown impossible to meetthe requirements in the 4 kc position.

4.2.4.2 Theoretical Considerations

The number of transformers needed, along with physical constants foreach, are determined by the specifications for BW, shape factor, andenvelope delay. According to universal curves for shape factor vs numberof tuned circuits and coupling, it would require about 30 criticallycoupled IF transformers to yield a shape factor of 1.9 shown in Table II.

On the other hand it requires only four transformers with a couplingfactor of three times critical to produce the same shape factor. Thishas the disadvantage of requiring at least one transformer to flatten outthe humps in the selectivity curve caused by over-coupling. After con-sidering various combinations, it was decided that the most economicaldesign would be five transformers. If they were all critically coupled, itwould yield a shape factor of 2.6 which is the requirement in the 4 kcposition. For the maximum BW position, four of these transformers could beovercoupled at about 2.5 times critical and the remaining one used to fill inthe double humps. This would yield a shape factor of about 1.8 with areasonable hope that the top of the overall selectivity curve could be madeflat with careful design.

It was discovered that the maximum linearity in the phase shift curveof a double-tuned transformer occurs at about 0.6 times critical couplingl.

1Gardiner and Maynard, I-F System Design Proc. IRE, VOL 32, P. 674, Nov. 1944

22


RESTRICTED

Four transformers at 0.6 times critical and one transformer at criticalcoupling gave the best envelope delay characteristic. This also yields areasonable design for the 2 kc position, giving a shape factor of about3·6. It should be noted that the best envelope delay characteristic wouldoccur with the minimum number of tuned circuits. Therefore, this designis a compromise between selectivity and delay.

The 0.1 and 1 kc positions were logically suited only to a crystalfilter. Even a 1 kc position using 5 transformers, undercoupled, wouldrequire Qs on the order of 200. Design of the crystal filter is includedlater.

The tuning capacitor was arbitrarily set at 1000 uuf. This is con-siderably higher than the usual design. However, since five or more trans-formers must be used for required selectivity, gain was not a problem andlow impedance tuned circuits could be used. This is desirable for maximumstability.

The center frequency is quite important in the design of IF trans-formers. The frequency of 455 kc was chosen for several reasons. One isthat it is considered standard in the industry for the final fixed IFfrequency. Another reason is based on bandwidth requirements. A lowerfrequency would make narrow bandwidths possible with reasonable Q values,but would make the wide position difficult. A higher frequency would workthe other way. A reasonable compromise is 455 kc for this particularspecification, Of course, IF image rejection and spurious responseentered into the selection also.

4.2.4.3 Overall Design

It has been established by theoretical considerations that the optimumnumber of transformers is five for the requirements shown in Table II. Howto arrange these transformers in a practical circuit and calculate theirdesign constants was the next problem.

Non-crystal Positions

The four non-crystal BW positions will be considered first. Thedesign was based on Gardiner and Maynard'a universal design curves fortuned coupled circuits The 4 kc position was obtained with five trans-formers, critically coupled with Q factors of about 125. It is assumedthat all transformers are isolated by tubes.

The 2 kc position was more complicated because of phase delay con-siderations. Using the fact that maximum linearity in phase shift occursat 0.6 times critical coupling, a design could be made. Five transformersat 0.6 times critical did not satisfy the requirements. It was necessary toadd one with a compensating phase delay characteristic. It was found that onetransformer at critical coupling just compensated for four

1Gardiner and Maynard, I-F System Design Proc. IRE, VOL 32, P. 674, Nov. 1944

23


RESTRICTED

transformers at 0.6 of critical. The design requires close tolerances tomeet the requirements. The 2 kc BW specification was also satisfied bythis combination using all Q factors of 125.

It should be mentioned that the amount of envelope phase delay isdetermined by the slope of the phase shift curve of the five IF trans-formers. The phase shift must be calculated to at least four decimalplaces in order to determine the change in slope that would cause the 4.63microsecond delay variation over the top bandpass of the five IF trans-formers. This sort of calculation is impractical by conventional means.

McCoy2, in his work on envelope phase delay, develops as a side-linean easy method of computation for attenuation as well as phase shiftcurves. This method is based on the exact equivalence, as far as se-lectivity and phase shift are concerned, of a series of symmetricallystagger-tuned single-tuned circuits to a series of double-tuned trans-formers having the same number of tuned circuits. This method is shownmathematically correct for any value of coupling in the transformers.

The big advantage of this method of calculation is in being able tofind quickly and accurately any desired selectivity curve or phase shiftcurve by simple addition and subtraction. By McCoy's method, it waspossible to calculate the performance of the proposed IF design before-hand. Actually, equipment was not available to measure envelope delayuntil the design was finished and transformers constructed. However,actual measurements were very close to predicted performance.

In the 8 and 16 kc BW positions, it was necessary to over-couplefour of the IF transformers as well as lower Q to obtain the desiredshape factor and BW. The first transformer is used to counteract therise in voltage on either side of the resonance caused by over-coupling.In order to design the fill transformer adequately, it was necessary tocalculate the attenuation curves over the top bandpass very carefully.The method of McCoy was used.

After the composite curve for the four over-coupled transformerswas obtained, a graphical method was used to find the optimum couplingand Q for the fill transformer to give the flattest overall curve. Inthis method, selectivity curves are plotted on log paper, and a compari-son which is independent of Q values is made by moving curves along thefrequency axis. After a suitable match of shape is found, i.e., thevalley matched to top of the fill transformer, then the relative Q re-lation is determined by how far one curve has been moved in relationto the other. In this manner, a satisfactory design was established.

2R.E. McCoy Envelope Delay in D-F receiversTechnical Memorandum M-1009, Sept. 20, 1946

24


RESTRICTED

The following chart gives the final design constants for the fiveIF transformers:

TABLE III

2 kc 4 kc 8 kc 16 kcTraf. Q K/Kc Q K/Kc Q K/Kc Q K/Kc

T-l 125 0.9 125 0.9 87 0.5 60 0.5

T-2 125 0.6 125 1.0 88 2.0 68 2.2

T-3 125 0.6 125 1.0 88 2.0 68 2.2

T-4 125 0.6 125 1.0 88 2.0 68 2.2

T-5 125 0.6 125 1.0 88 2.0 68 2.2

T-6 26 0.9 26 0.9 26 0.9 26 0.9

Q = wL/R for both primary and secondary

K/Kc = KQ = MQ/L

M = mutual inductance

L = inductance of primary and secondary

Notes: (1) Final transformer have unequal Qs. However this chart gives the equivalent equal Q transformer.

(2) Transformer T-6 above is designed to feed the diode load, while not appreciably attenuating the top bandpass in the 16 kc position.

(3) Transformers remain in the 2 kc position for the 0.1 and 1.0 kc crystal filter positions.

Crystal Filter

The crystal filter and associated circuit was designed by Bob Craiglowof Collins Radio Company. It is a conventional design except for themethod used to obtain sharpest bandwidth. Instead of the crystal workinginto a low resistance load, it feeds into a capacitor. By this means, thefull Q of the crystal is realized, and at the same time, a reasonable out-put is obtained. The crystal frequency is pulled slightly, but not enoughto cause trouble. The 1 kc position was obtained by adding additional re-sistance in series with the crystal to lower its Q.

25


RESTRICTED

Circuit Considerations

The actual circuit layout with tubes, resistors, etc. was designedmostly on the basis of laboratory experiment. It was found that the filltransformer, T-l, should be placed first in the line up. Otherwise therise in gain on each side of resonance (about 9 db) could overload thelast stages before the fill transformer. Also the change in gain between4 kc (max. gain) and 16 kc(min.) positions had to be compensated. Thiswas accomplished by switching cathode resistors in two stages of ampli-fication.

Amplifier layout was conventional with the usual number of decouplingresistors and capacitors. High quality components were used throughout.Size of bypass capacitors was more than adequate. No trouble with re-generation was encountered as a result of good isolation between stagesand between input and output. Several tubes were tried for 455 kc ampli-fier use. Final tube decided upon was a 6BJ6, a remote cutoff pentodewith 150 ma 6v filament. The last IF amplifier tube was chosen to be a 6AK6.This tube provided the large signal input capacity necessary to meet aspecification calling for the output to be linear with 20 db signal increasefrom normal level, MGC position.

In Figure 57 is a simple block diagram of the main components.

Overall gain required was determined by needs of RF section andaudio. Final value was 200 uv input for 7 vdc on detector. Gain couldbe varied by means of gain adjust pot between 10,000 and 100,000. As canbe seen on the block diagram, all tubes serve as both amplifiers and gaincontrol devices except the final amplifier (6AK6). This would make itdifficult to eliminate one or more tubes by capacity coupling betweentransformers. Nevertheless, it was found experimentally that 22 uuf topcapacity coupling between transformers would be satisfactory (no ap-preciable change in selectivity) if a resulting 20 db loss in gain couldbe recovered.

An external IF output is provided for teletype use. Nominal im-pedance is 60 ohms and normal output is around 30 millivolts. This isprovided by a cathode follower which is fed from the grid of the finalIF amplifier. The BFO is effectively isolated from this output by thefinal IF tube (6AK6). Output voltage must be linear from 20 to 200 mvout. The cathode follower tube (1/2 12AU7) must operate near maximumratings to obtain 0.2 volts at 60 ohms impedance.

4.2.4.4 Transformer Design

A previous Table gives the design constants for the five IF trans-formers. Some method of variable coupling had to be found which wouldbe adjustable from the front panel and yield the desired change incoupling. The following table gives the range of coefficient of couplingK involved for each transformer.

26


RESTRICTED

TABLE IV

2 kc 4 kc 8 kc 16 kc Traf. K K K K __

T-l .0072 .0072 .00575 .0084

T-2 .0048 .008 .0228 .0324

T-3 .0048 .008 .0228 .0324

T-4 .0048 .008 .0228 .0324

T-5 .0048 .008 .0228 .0324

T-6 .0345 .0345 .0345 .0345

___M___ K = √√(L1 L2)

M = mutual inductance

L1 = inductance primary

L2 = inductance secondary

Note: These values are again based on the equivalent equal -Q transformer.

Variable Coupling Methods

Several methods of variable coupling were investigated both theoreti-cally and practically. A ratio of at least 7/1 was needed between maximumand minimum coupling, as shown in Table IV. Remember that this refers tothe actual physical coupling, which is independent of Q values.

Top capacity coupling is used sometimes to add to the inductivecoupling between IF coils. It has the advantage of simplicity and ease ofapplication. However, it has two serious drawbacks which prevent its usehere. One is that if top capacity is used to provide most of the couplingat high coupling factors, the selectivity curve will be inherently asym-metricall. The other defect is that in switching between several capaci-tors to provide variable coupling, the coils are detuned with each change

1 Raditron Designer's Handbook, 3rd Edition, P. 125.

27


RESTRICTED

in top capacity. This last defect is present in several variable couplingsystems, and prevents their use. Also, high impedance switching, which wouldbe necessary, presents a design problem in high gain amplifiers. Carefulplacement of leads and switches must be observed or regeneration trouble willbe encountered.

Bottom resistance or reactance coupling was investigated also. Thismeans placing a common impedance in the cold end of primary and secondarycoils. The common reactance method has the same drawbacks as top capacitycoupling, except that high impedance switching is avoided. Bottom resistancecoupling was investigated experimentally, since little could be found in theliterature. It was found impossible to obtain even critical coupling withthis method. Evidently the common resistance lowered coil Q faster than itprovided tighter coupling.

The most basic method of varying the coupling between two coils ismoving them physically in relation to each other. This method is the mostdirect and easiest to understand of all the schemes. Where a powdered ironcore is used, the whole coil and core assembly must be moved together inorder to avoid the serious defect of detuning. Even then, it is hard toavoid some change in tuning because of stray capacity to shield can, spacers,etc.

Two possible drawbacks are found in this system. Mechanical constructionis complicated and to meet the particular requirement of four distinctcouplingpositions would require an accurate cam and detent arrangement.This is difficult, but could be done. Perhaps the more serious objectionis difficulty in obtaining the 7/1 coupling change required for this design.One such transformer constructed experimentally had a coupling change ratioof 3/1, which would still not be sufficient for the requirements.

Another possible method is variable stagger-tuning. Stagger-tunedcoils could be arranged to give the same overall selectivity requirements.Selectivity could be changed by switching fixed capacitors across thestaggered pairs. This method was not investigated because of complicatedcalculations and the need for isolation between coils (more tubes). Indi-cations were that this method would be better suited to high frequency IFstrips.

The first method tried and actually used for a complete set of trans-formers, was a tapped link arrangement. A sketch of the transformer isshown in Figure 58.

In the diagram, L1 and 12 are iron core tuned coils with powdered ironcups around each coil. Lx is several turns of wire wound inside the ironcup for L1. Lx and L1 are coupled tightly. L1 and L2 have enough couplingto provide the first of four coupling positions.

28


RESTRICTED

This construction and method has several distinct advantages. Theswitching is done on the cold end of the secondary only. This makes layoutuncritical. Series resistors R1 and R2 can easily be inserted to lower theQ, and mechanical construction is straightforward. The only critical itemis where the taps are placed on Lx. This determines the change in couplingbetween positions 1, 2, 3 and 4. A relatively large range of coupling canbe obtained by adding more turns to Lx.

The big disadvantage, and what finally disqualified the system, waschange in tuning of the secondary by switching more turns in series with L2.This causes the double humps in the over-coupled positions to be lopsided.In one transformer the effect is not too pronounced, but when four or fiveare added together, the effect is so great as to make it impossible to obtaina flat topped overall response curve.

Various systems were tried to compensate for detuning. One was toswitch in a series condenser for every new coupling position to compensatefor the change in inductance. This was successful, but the added com-ponents made it somewhat impractical.

The system finally used was an outgrowth of the tapped link inductance.To change coupling, it was decided to change the coupling of Lx in relationto L1 by changing the spacing. Lx was placed outside the iron cup sur-rounding L1, and moved up and down the coil form to vary coupling, whilekeeping the inductance of Lx constant. Several practical difficulties wereencountered, but overcome in the final version.

Physical Construction of Transformer

The interstage transformer shown in Figure 60 having four separatevalues of coupling will be discussed first. T-l and T-6 in Table IV werean outgrowth of T-2 thru T-5 (interstage transformers).

The coupling coil, Lx, discussed in the last section is now coils A,B, C, and D, Figure 3. Each coil has the same inductance, but varies inits coupling to L1 due to its physical spacing. Each link corresponds toan overall coupling factor for the transformer. The primary tuning coregoes through all coils A, B, C, and D to provide higher coupling to primaryL1. Note that this core is connected to its adjusting screw through aphosphor-bronze wire. This wire causes little loss in Q in the primary.A brass stud would de-Q the primary considerably.

The powdered iron half-shells serve several purposes. One is to in-crease the inductance of L1 and L2, which means that fewer turns are re-quired. They shield the coils from external fields. This is quite im-portant, because in the completed IF strip, the transformers are mountedas close as possible to each other to save space. The primary iron cupserves to shield the links from the secondary.

29


RESTRICTED

Wire used for the main coils is No. 30 x 44 SSE Litz, This gives ahigh Q for the relatively few turns used (83). Small links are 8 turns ofNo. 30 DE wire. Powdered iron parts are all carbonyl E basic material.This is a high Q, low permeability iron. Phenolic parts are made of thebest electrical grade of paper base pehnolic available at the time.Collins Specification No. 278 0103 00 describes the mechanical and electri-cal specifications in more detail than will be covered here.

The original design called for the main coils to be mounted coaxially.This had the disadvantage of having one tuning adjustment below the chassis.Shield can size on the first transformer was 2-1/2 inches high by 29/32inch square. To enable both tuning adjustments to be made from the top,the construction of Figure 3 was used. Shield can size is 2-1/2 incheshigh by 1-7/16 inches square.

Coupling System

The final variable coupling system will now be explained in more de-tail. A system that will give four positions of Q and K/Kc, as listedunder Table III, transformer T-2, is the goal. Figure 58B shows the cir-cuit for the transformer.

It was decided that for simplicity in switching, it would be betterto lower the Q of the secondary only for the wide positions. This can bedone easily by inserting resistors in series with the secondary tank cir-cuit as shown in Figure 58B. The only requirement is that these resistorsbe non-inductive. The overall switch arrangement in Figure 58B is welladapted to the tandem switching of five transformers. All leads are nearground potential and are low impedance. No trouble was encountered withcoupling between switch sections and the switch shaft, even though about10 inches long, caused no trouble with coupling between input and outputtransformer.

The theory of designing an unequal Q transformer to be equivalent inselectivity and coupling to the equal Q design shown in Table III, wasdeveloped by J. E. Maynard l. Applying his methods to the transformer inquestion yields the following design:

TABLE V

Interstage Transformer T-2 Thru T-5

Factor 2 kc 4 kc 8 kc 16 kc_ Q pri 125 125 125 125 Q sec 125 125 68 46.7 K .0048 .008 .0235 .033 K/Kc .60 1.0 2.1 2.5 ___1___ Kc = √√(Qp Qs)

1 J.E. Maynard, Tuned Transformers. G.E. Review VOL 46, P. 559, Oct., 1943

30


RESTRICTED

This transformer will have the same characteristics as transformerT-2 listed in Table III. It was found experimentally and by theory thatunequal Q in primary and secondary produces no asymmetry in the responsecurves. The response is exactly the same in all respects as the equivalentequal Q transformer. Transformers can be checked on a Q meter or in anamplifier circuit for conformance with Q and coupling, providing that thevalues listed are used. K/Kc listed in this table is the quantity com-monly called N or C, the ratio of actual coupling to critical coupling,that is measured by the following relation: ___________ N = K/Kc = √√((Eo/Et)-1)

where Eo = primary RF volts with secondary open

Et = primary RF volts with secondary tuned

It was found also that a transformer with unequal Qs will have ahigher stage gain than the equivalent selectivity equal -Q transformer. Inthe schematic of Figure 58B, the secondary output must be taken across thecoil L2 to yield a symmetrical selectivity curve. If the output voltageincludes the link voltage, i.e., output across C2, then the selectivitycurve will be asymmetrical especially at high coupling. Considerable work wasdone during the development on the unequal Q theory reviewed briefly in thepreceding paragraphs.

It was found experimentally that there was not quite enough roombetween the primary and base board to piece four links covering the couplingrange involved. A solution was found in reversing the polarity of the 2 kclink C. This caused its coupling to oppose the main coil coupling and re-sulted in the desired overall K factor. In fact, it was found possible toobtain exactly zero coupling by moving link C about 1/16 inch further fromprimary than its present position. The range of coupling available thenby this method is practically unlimited, depending on how many turns areon the links and whether they aid or oppose main coil coupling.

The particular coupling setup used was decided upon by experiment andby trying different combinations on paper. First of all, the links had tobe spaced far enough from each other so that adjustment could be madewithout interference. The coupling desired between the two main coils hadto be physically obtainable. Much work was done with an experimentaltransformer in which everything was movable. The optimum number of turnson the links turned out to be 1/10 of the main coil turns, i.e., 8 vs 83turns.

The coupling between primary and secondary with no link in seriesworked out to be .0040. This is the actual coefficient of coupling _ _M__K =√√(L1L2). The additional coupling provided by links A, B, C, and D, either

31


RESTRICTED

aids or opposes this original coupling. The following table lists variousvalues of Kx shown in Figure 58A and the resultant overall coupling

TABLE VI Interstage Transformer

2 kc 4 kc 8 kc 16 kc

Main coil coupling, K12 .004 .004 .004 .004

Link coupling Kx -.0088 +.004 +.0195 +.029

Resultant overall -.0048 .008 ·.0235 .033 coupling K ______K12 = M12/√√(L1L2)

______Kx = Mx/√√(L1Lx)

The last line agrees with the desired coupling shown in Table V. Theminus sign has no significance in the final answer. It merely indicatesthat Kx opposes K12 in the 2 kc position. This table shove that link D

(4kc) is the farthest from primary coil L1. The fact that couplings areadded and subtracted in this manner was shown experimentally.

Each coil, A, B, C, and D in Figure 58A has the same inductance.One-half turn was removed from coil A to keep its inductance the same.This was because it was under the iron half-shell, which tended to in-crease the inductance. The primary timing core, which goes through allfour links, helps to keep their inductance constant regardless of position.

Other Coils in IF Amplifier

The input transformer, T-l, of Table III and IV is the same as theinterstage transformer in physical construction. However, the linkspacings are varied somewhat along with Q values. Refer to CollinsSpecification No. 278 0102 00 for physical construction and schematic.It has three positions of coupling, and hence only three links. Thefollowing is design data for the input, transformer T-l:

32


RESTRICTED

TABLE VII

Input Transformer T-1

Factor 2 kc 4 kc 8 kc 16 kc

Q pri 125 125 125 125

Q sec 125 125 67 39

K .0072 .0072 .006 .012

K/Kc .9 .9 .55 .8

K12 .004 .004 .004 .004

Kx +.0032 +.0032 -.010 -.016

K overall .0072 .0072 .006 .012

Notice again that the link coupling in the 8 and 16 kc positionsopposes the main coil coupling K12. This is necessary because the 4 and8 kc overall coupling are so close together. The 8 kc link is reversedin polarity in order to separate it physically from the 4 kc link. Thereis one peculiar effect resulting from reversing the link polarity. If themain coils are moved closer together, the effect is to increase the 4 kcoverall coupling as would be expected. However, the 8 and 16 kc overallcouplings are decreased by this operation. This can be understood bylooking at the values of K12 and Kx in Table VII.

The output transformer is operated at one value of coupling and Q.Mechanical construction is the same as T-1 through T-5. The relativelyhigh physical coupling, K = .0345, could be obtained in this design onlyby a link placed as in T-2, link A, the interstage transformer. Here themain coil turns equal 125 and link turns are 15. This transformer oper-ates into a series diode load of 22K ohms.

The AGC amplifier and crystal filter circuits both use a high im-pedance single-tuned circuit. Here the iron half-shells are mounted openends together to completely enclose the coil. The crystal filter unit isconstructed with the crystal phasing control (5-12 uuf) and loading coilmounted in the same shield can.

There is a 455 kc transformer mounted on the RF sub-assembly unitwhich feeds into the crystal filter. This transformer presented a specialdesign problem. The bandwidth had to be broad enough so that it did notattenuate the 16 kc selectivity curve appreciably. This required a Q ofnot more than 12 to 15 under operating conditions. The output coil wasto be a center-tapped untuned link of about 1000 ohms impedance. Any

33


RESTRICTED

higher impedance would load the crystal filter too much, since this im-pedance is effectively in aeries with the crystal. A lower impedancewould result in not enough gain from the last mixer grid to the first IFgrid. The resulting design was a transformer with more turns on the out-put link than on the main coil. This is just the opposite from a con-ventional transformer used in this application.

4.2.4.5 Practical Considerations

Transformers

From the start of the transformer design, it was realized that tuninghad to be done in one BW position only, and that the selectivity curve hadto stay on frequency and remain symmetrical when switching to any otherposition, otherwise, the receiver would have to be retuned between BW po-sitions, and the signal might be lost in going from a wide position to asharper one to eliminate interfering signals. This requirement meant thatthe tuning of the secondary of switched transformers had to stay right onfrequency.

A further requirement was that each transformer could be tuned bysimply peaking each iron core for maximum output voltage without the needof special tools or elaborate techniques. T-1 is tuned in the 8 kc overallposition. This corresponds to its sharpest position. T-2 through T-5 aretuned in the 2 kc overall position. The output coil, T-6, is somewhat hardto tune because of low Q and high coupling. However, it can be tuned afterseveral operations between primary and secondary cores. Tuning procedurefor the crystal filter and AGC coil are also quite simple but require atunable signal generator for setting the phasing condenser.

Electrical tolerances are covered in detail on the respective trans-former specifications. The need for the relatively narrow tolerances of ±±5%and ±±10% is based on several factors:

(a) Overall requirement for bandpass phase characteristic.

(b) Change in selectivity curve symmetry with JW position.

(c) Filling in the overcoupled humps to produce a flat selectivity curve top.

(d) Original specification for bandwidth tolerance.

It has been determined both on paper and experimentally that if theindividual transformer specifications are satisfied in each case, then theoverall IF characteristics will be within specifications. Each individualspecification has a short production testing procedure, which will insurethat the transformer will operate satisfactorily in the complete unit.

34


RESTRICTED

Electrically the important quantities are bandwidth and coupling. Theyare particularly important in the direction finding position (2 kc) and inthe over-coupled positions. For this reason, BW and coupling should bechecked for production testing. Q and gain are not the important quantitiesin this case, and can only be checked as a type test.

Mechanical tolerances are entirely determined by the electrical speci-fications. It was found that if link spacings to primary coil are held with-in 1/64 inch, the electrical requirements could be met. Also spacing betweenmain coils should be held to 1/32 inch and distance between coil form to.005 inch. Turn on links A, B, C, and D must be held at least 1/4 turn.Permeability of primary tuning core must be held to ±±1%.

Overall Circuit

Some trouble was encountered with the switches used to change band-width. After being used for several months, or particularly after a humi-dity run, the switches would show enough contact resistance to cause erraticoperation. Resistance on the order of one-tenth ohm will be enough to changethe secondary Q 5 to 10%. Switches with double contacts (front and rear inparallel) were finally used. Rotor is solid coil silver and clips are silveralloy. No more trouble was encountered with this type construction.

With six amplifier stages in tandem, it was necessary to control thegain from set to set by using a variable potentiometer in one cathodecurcuit. This pot is adjusted to the correct overall gain at the time IFstrip is first tested. Another problem was the change in gain between BWpositions. This is compensated for by switching cathode resistors commonto two stages. However, since each transformer can vary ±±15% in its gain,the gain between bandwidth positions could vary 7 to 8 db or more. This isprovided for by possible selection of individual gain control resistors, ifnecessary, in final test of the IF strip.

There is a nominal 2700 ohm resistor in series with the 455 kc crystal.This can be changed in test to adjust the bandwidth in the 100 cps position.This was necessary because of crystal activity (Q) variation from differentmanufacturers.

Summary of Measurements

The last three complete IF units built up to September 23, 1952, weremeasured for performance according to a standard procedure. Data was re-corded on uniform data sheets for convenience in comparing units.

In the first unit, the 0.1, 2 and 16 kc bandwidths were slightlywide. In the second, only the 2 kc BW was slightly wide. In the lastunit, all positions were within specifications. Windings were dimensionedaccording to a standard when given to the coil winding department. Peak-to-valley ratios were all well within specifications.

35


RESTRICTED

BFO range was slightly off on one unit. This was corrected. Inter-modulation distortion was out of specifications on all units. The speci-fication figures were based on the first IF strip built. Several changeswere made in tube operating conditions since that time which evidentlycaused this characteristic to be slightly worse. Direction findingcharacteristic was out of specification in all units. However, it wasdecided that the 1/4 degree tolerance was not possible. All units met±±l deg tolerance which the Signal Corp indicated might be acceptable.

BFO leakage at 60 ohm IF output was out of specification if a changein frequency of ±±3.5 kc was included. Neutralization could be made atone frequency, but did not hold very well over the range of frequenciesinvolved. This specification should be investigated, especially as to methodof measurement. All other items measured were within specification on thethree units.

4.2.4.6 Service Conditions

Impregnation Against Humidity

Protection of high Q, high impedance coils against humidity isalways a problem, particularly in an IF transformer where coils consistof many turns built up into a large pi. In this case, coils had tomeet a 48-hour soak at 95% relative humidity at temperatures from 20 degC to 65 deg C. Measurements were to be made with coils actually inhumidity chamber. No numbers were written into this specification, butit was decided to try for the best possible and practical method forprotection.

The problem is to keep moisture out of the coil. If it penetrates,the Q of the coil drops as well as a change in tuning. This is due toone or all of the following factors:

1. Dielectric losses of water.

2. Change in tuning because of change in dielectric constant of water over air, or change in permeability of iron core.

3. Lowering of impedance between strands of Litz wire and between adjacent turns because of moisture penetration through varnish on wire, or through contact at pinholes, etc.

4. Eddy current losses in water.

The third item was demonstrated by separating the strands of 30 strandLitz wire into two groups of 15 strands each. A DC resistance measurementwas made between these two sections for testing. At room conditions theresistance was on the order of 100 megohms. Immediately upon dipping aninsulated section of wire into water, the resistance would drop to a

36


RESTRICTED

magnitude of 100 K to 1 meg. It would then continue dropping slowly. .Thenew nylon-enamel wire (Dipsol, etc.) was very much worse in this test.Even teflon covered (Ceroc) wire was very poor. This last was probably dueto cracks or pinholes in the insulation.

The following commercial preparations were investigated for possibleuse as impregnants. Those that were actually tried are indicated by anasterisk.

1. Cycleweld 55-9 cement, Chrysler Corp.

2. EC870, polychloroprene base compound 3-M Co.

*3· 3-M casting resin R-1433.

*4. Durez 14139, phenolic thermosetting resin.

*5. Q-322 wax, Zophar Mills.

*6. C-770 wax, Zophar Mills.

*7. Plaskon ST-856, silicone alkyd resin, Libby-Owens-Ford Glass Co.

*8. MR3738, potting wax, Mitchell-Rand Insulation Co.

*9. MR3784, dip coating wax, Mitchell-Rand Insulation Co.

10. Acme Star Compound, Acme Wire Co.

11. Penacolite Adhesives, Koppers Co., Inc.

12. NBS casting resins.

*13. DC935 silicone resin, Dow Corning Corp.

*14. Araldite Casting Resins B, D, F, G, Ciba Company.

*15. Araldite adhesives 11, 15, type 1,101, Clba Co.

16. DC996 silicone resin, Dow Corning.

The waxes and the casting resins both worked very well in protectingagainst moisture. All the waxes were for high temperature use. However,since they are thermoplastic, they will always melt or run at a hightemperature. The highest temperature wax tried was MR3738, which requiredapplication at 160 - 180 deg C. Waxes all gave very good penetration.

Complete details on how the coils were tested would be too lenthly toinclude here. The whole subject of impregnation would require a separatereport. Of all the compounds tested, Araldite Resin F, now called CN-503,had the greatest number of desirable qualities.

37


RESTRICTED

1. Thermosetting - will not melt or run under any temperature.

2. Temperature range outstanding (no cracking at -62 deg C).

3· Simple application by hot dipping.

4. Extreme fluidity - viscosity of water with a wetting agent when applied.

5. Excellent penetration without vacuum techniques.

6. Outstanding adhesion to metals and plastics.

7· Gluing and impregnation in one operation.

8. Very good resistance to water and moisture.

9. Hardly any shrinkage with curing (0.5 to 2%).

10. 100% solids. No solvents.

11. Coil form inside can be dipped without appreciable build up of resin.

12. Iron cores can be impregnated with minimum increase in diameter.

Disadvantages:

1. Two compounds to be mixed before using.

2. Long cure at high temperature (10 hours at 130 deg C or 48 hours at 100 deg C)·

3· Pot life of resin is limited because of thermosetting properties (6 hours at 120 deg C).

4. Parts must be hot dipped (above 60 deg C). This is an advantage in some cases.

The last three IF unite had coils impregnated with Araldite CN5O3.Coils were tested under humidity by operating them as a complete IF ampli-fier on a special test chassis, the same as is used in the complete re-ceiver. The following table summarizes the results of testing. Gainmeasurements were made immediately upon removing chassis from test chamber.Figures represent microvolt input level to all six stages in tandem forstandard output.

38


RESTRICTED

Input Humidity Input Input After IF Unit (Before Cycle) Cycle (After Cycle) Returning

R390 #4 2 kc 95 6 days 250 210 16 kc 57 SC-D-159l4 83 80

R389 #3 2 kc 55 48 hrs 400 150 16 kc 30 JAN-R-93 47 33

R389 #3 2 kc 55 65 hrs 150 140 16 kc 30 JAN-M-745 36 36

R390 #4 2 kc 22 4 days 150 65 16 kc 21 SC-D-159l4 41 35

R390 #4 2 kc 52 4 days 260 92 Primary cores 16 kc 44 JAN-R-93 43 43 impregnated

R390 #4, both 2 kc 52 4 days 78 70 primary & sec. 16 kc 41 JAN-R-93 46 43 impregnated

The amount of detuning of transformers was found mostly due to changein iron core permeability (after cells were impregnated). This was shownby impregnating half the iron cores in the unit. After humidity run, thecores that were impregnated needed very little adjustment for tuning. Coreswith nothing for protection, were consistently off tune by one or two db.This is not much per stage, but with six stages, the effect is quitenoticeable.

Vibration and Temperature

Complete IF unit performed satisfactorily under vibration testing.The resin used for impregnating provides an excellent supporting agent forcoils and powdered iron cups. It is flexible enough not to break undershock and vibration if leads are not too short between gluing points. Atleast two 6BJ6 tubes failed on the package tester machine and burned outthe stage decoupling resistor. This should be checked if it happens again.

Temperature tests were conducted on the last three IF units. Datashoved that the center frequency in the 2 kc position shifted about 1.5 kcfrom -40 to +60 deg C, with a reduction in gain of about 8 db. This ismaximum change of the transformers in the sharpest position. The 1 kcshift in peak frequency was about 0.75 kc over the same range with a re-duction in gain of the same magnitude. The 0.1 kc position peak shiftedabout 150 cps over -40 to +60 deg C. Gain again changed about 6 db. Thiswas with no temperature compensation in IF transformers.

39


RESTRICTED

Production specifications on each transformer call for less frequencyshift than transformers made so far have shown. This would requiretemperature compensating condensers and should result in the overall fre-quency shift being about half the shift shown above.

Production Testing

Testing Each Transformer

The specification for each transformer calls for production testing bychecking the bandwidth at one attenuation point and coupling. The mostaccurate way to check these quantities is with the conventional signal gener-ator, frequency meter, and output meter method. However, this method doesnot lend itself well to high speed production testing.

Another method which is not generally as reliable, makes use of acalibrated sweep generator and CRO for observing the selectivity curve.It is hard to measure actual bandwidths this way, and next to impossibleto measure coupling of an undercoupled transformer. However, as soon asthe transformer develops double humps by overcoupling, the measurement be-comes feasible. The peak-to-valley ratio is a very accurate method ofmeasuring relative coupling. The distance between humps, which is usuallyquite distinct, gives the bandwidth required. These two quantities, peak-to-valley ratio and bandwidth between peaks, describe the transformer veryaccurately, and can be the basis for close control.

The catch here is that a transformer that is undercoupled cannot bechecked accurately. It can be shown, and is discussed by Termanl that thevoltage versus frequency curve across the primary side of a double tunedtransformer exhibits double humps before the secondary side. In fact, thetransition between one hump and two humps occurs at a relative coupling ofabout 0·55 times critical. If a cathode follower is fed with the primaryvoltage and the output of the follower rectified and fed to a CRO, theresulting selectivity curve will exhibit just such action.

This setup was tried experimentally with success. The interstagetransformer with 0.6 times critical in the 2 kc position showed doublehumps on the primary aide. To establish what the peak-to-valley ratioand bandwidth between peaks should be on the primary side, it is necessaryto measure the transformer conventionally for compliance with specifi-cation, then measure the primary selectivity curve.

The following setup was tried for a possible production test setup:A sweep generator was fed to two transformer test circuits in parallel.One transformer was the standard and the other under test. The outputs

1 Terman, F.E. Radio Engineering, 3rd Edition, P. 58.

40


RESTRICTED

were rectified for DC and fed to an electronic switch. The output ofthe switch was observed on a scope. In this manner the calibration is de-pendent only on the standard transformer and relative channel gain. It isindependent of sweep width, frequency or slight non-linearities that canoccur in such a system. The gains are adjusted to be the same through eachchannel and the two curves compared directly as to peak-to-valley ratioand bandwidth between peaks. If necessary, output is taken from the primaryside of the transformer to obtain double humps. With this method, changeson the order of ±±5% in BW and coupling can be observed easily. However, ameans for establishing maximum and minimum limits still is a problem. Onesolution is to modulate the standard curve at about 50 kc. This effectivelywidens it vertically and can serve as a sort of tolerance limit. It isfelt that this method is a solution to the problem of production testingIF transformers.

Testing Complete Unit

Production testing of the complete IF amplifier has not been decidedupon. Complete engineering tests for each unit have been set up and a datasheet made. These include service conditions tests. The last three IFunits have had complete data taken.

4.2.5 R-389 RF

During the planning stage of this contract, the general physical layoutof both the R-389 and R-390 had been determined. This resulted in adefinite limit as to the size and shape of the RF sections and since theR-390 portion was about two months ahead of the R-389 in development itmeant that little or no compromise in size could be made. Later in theprogram it appeared to be somewhat of a handicap to make the two receiversso similar in layout, but at this point the R-389 RF had to be designedinto the prescribed space.

The primary determining design factor was the specified 455 kc IFamplifier. This forced several immediate conclusions -

1. The circuit must be a multiple superheterodyne.

2. The RF tuning range would pass through the IF frequency.

The very accurate tuning and stability requirements imposed manyrestrictions on the electrical and mechanical design. The major mechanicalproblems in this category were providing a frequency indicator of sufficientaccuracy, and coupling the frequency indicator with a sufficiently smallamount of backlash to the variable frequency oscillator.

Electrical stability was determined by the oscillators to be used.The VFO was required to provide linearity as well as stability, and anyother oscillators used in conversion would be required to meet stabilityspecifications.

41


RESTRICTED

The need for tuning through the IF frequency (455 kc) had determinedthe necessity of a double superheterodyne. In addition adequate shieldingwould be necessary. Reduction of spurious and image would also require afirst IF frequency several times greater than the highest signal frequency.

4.2.5.1 The Signal Path

The initial considerations had pointed to the need for a double super-heterodyne. The signal path would then be as shown in Figure 61A.

In most receiver designs, RF selectivity would be designed for maxi-mum. However, due to the low frequencies involved, and the relativelywide pass bands required, it was necessary to design the R-389 R-F cir-cuits a pass band rather than for maximum selectivity (i.e. 8 kc pass bandat 115 kc.).

To further control the spurious response, it was desired to limit themaximum RF voltage on the first mixer grid under strong signal conditions.Two RF amplifier tubes with AGC were used for this purpose. One tube wouldhave been sufficient to provide the desired signal-to-noise ratio.

The number of RF coils used was set at 4. This would provide anantenna coil, first RF plate, and a coupled circuit for the second RF plate.The double tuned circuit was placed in the plate circuit of the second RFamplifier to reduce the amplitude of harmonics generated in the tube andsubsequently coupled to the first mixer. This was a spurious responseconsideration.

The most obvious spurious response is produced by the signal fre-quency harmonics falling on the first IF frequency. Past experience hadshown that these responses should not be lower than 6th or 7th order ifadequate attenuation is to be obtained. The highest signal frequency is1500 kc and 7 x 1500 = 10,500. Thus, if a 10,000 kc IF were used thelowest order harmonic spurious would occur at 1428.6 kc and would be the7th order. Accordingly, the first IF was selected at 10 mc. This se-lected frequency scheme also limited cross over spurious to the 7thorder.

The injection frequencies were now determined. Injection to the lastmixer would be either 9545 kc or 10,455 kc. The first mixer injectionwould be 9985 kc for a 15 kc signal and 8500 kc RF signal.

The Injection Path

The primary problem in supplying the injection voltages lay in thestability requirement. The situation is aggrevated by the fact that bothinjections are several times higher in frequency than the signal.

The first injection frequency range is 8500 kc to 9985 kc. A variablefrequency oscillator operating directly on this frequency would fail toprovide either the desired calibration accuracy or stability. The secondinjection frequency stability would also be marginal even with the bestcrystal and ovens available.

42


RESTRICTED

It was decided that the stability problem could best be solved byproviding a ring circuit in which the drift of the second injection wouldbe completely canceled. This would also enable the variable frequencyoscillator to operate at a frequency nearer to the signal frequency. Sucha scheme placed the second injection frequency at 10,455 kc.

Early in the development it had been considered desirable to providea dial system presenting 100 kc segments up to 500 kc and then cover thebroadcast band in one sweep. This scheme would permit the use of a slide-rule type dial presentation. Without further comment the conversion systemproposed was as shown in Figure 61B.

This system provides an injection range at the first mixer of 10,000to 8400 kc which gives an RF input range of 0 kc to 1600 kc.

A breadboard model of this system was constructed. Spurious responsesdue to the low frequency crystal fundamental and harmonics was of a greatenough magnitude so that considerable improvement was in order. Due tothe pressure of time the system was reconsidered. Spurious responses werefurther examined (Reference Data for Radio Engineers, 3rd edition, pages455 - 458) and the conversion system modified to be as shown in Figure 62A.

This system was tried on the breadboard and by substituting a signalgenerator for the VFO, tests were made. The results were very encouraging.

A request for a six-band oscillator was presented to the oscillatordepartment. It was their opinion that while the oscillator could probablybe built, the predetermined space allotment and the limited time availablewould prohibit the project.

The good features of the frequency scheme proposed for use with thesix-band oscillator indicated that the basic system should be retained ifat all possible. This could be done if the decade dial system weredropped and a continuous system used whereby the oscillator would tune arange giving an RF input of 15 kc to 500 kc in one sweep. This could beaccomplished if the oscillator tuned from 470 kc to 955 kc. It wouldthen be possible to tune the broadcast range by using the second harmonicof the oscillator and extending the high range to 977·5 kc. To provideoverlap and have a tuning range of 14 kc to 1505 kc the oscillator shouldtune from 469 kc to 980 kc. The conversion scheme is as shown in Figure61B.

The 469 kc to 980 kc oscillator was developed under nomenclature70H1. This oscillator had solved the frequency scheme problem; however,additional design considerations were presented as a result. Tuningthe front end in two continuous sweeps - 15 kc - 500 kc and 500 kc -1500 kc - presented several problems.

43


RESTRICTED

4.2.5.2 Practical Design Considerations

The Signal Path

The decision to cover the frequency range in two sub-ranges pointedout the utility of a counter as a frequency indicating device. A counteris linear in presentation and so the variable frequency oscillator anddrive shaft to the RF section must have linear responses with respect tofrequency. Investigation led to the design of RF coils that would have aresonant frequency varying linearly with core insertion.

Without going into design details it may be said that the shortestpossible core travel that could be used to give approximately at 2 to 1frequency coverage was 1-1/4 inches. A 1-3/4 inch core was used. Coilswere designed 1-1/2 inches long. At higher frequencies a lesser travelcould be used due to the smaller effective diameter of solenoid coils.

It was shown at the start of the development that one serious shortcoming of conventional low frequency receivers lay in ringing duringtransient surge reception. It was felt this could be minimized if the Qand impedance of the RF coils were held to the minimum usable values.Computations considering the following led to the selection of values:

1. Minimum antenna gain necessary to obtain sensitivity.

2. Q versus bandwidth consideration.

3. Stage gain.

4. QX limitations as determined by physical coil size.

The QX was set at approximately 5000 ohms and the Q factor from 4·5 to 30. The lowest Q occurs at 115 kc and the highest at 1505 kc.

In designing individual coils three necessary factors were kept inmind:

1. Linear frequency versus core insertion. 2. Linear frequency versus Q relation.

3. Impedance.

The reason for the linear tuning has been mentioned before. The linearfrequency versus Q relation gives a constant gain over each band. The im-pedance value as determined by the 5000 ohm QX set the inductance. It isinteresting to note that the lowest frequency coil is resonated with a .02microfarad capacitor. The highest broadcast coil is tuned with 620 micro-microfarads.

44


RESTRICTED

Each of the RF coils is wound with pie windings in series, each of anumber of turns as determined by the desired linear curve. The lowestfrequency band coil contains 10 pies, all other coils have 12. Varioussizes of Litz and solid wire are used as determined by Q. Each coil istrimmed by a variable inductance trimmer. It was decided to provide atrimming range of ±±4% and the values of capacity required to do thiswould be prohibitive. In addition, a permeability tuned coil is moreeffectively trimmed by a series inductance trimmer than by a shunt capacitytrimmer . However, some loss of incremental permeability is incurred.

The complete frequency range was divided into seven bands as follows:

14 - 28 kc 250 - 505 kc

26 - 56 kc 495 - 870 kc

54 - 120 kc 865 - 1505 kc

115 - 245 kc

As it had been decided to use four RF coils per band, the total numberof RF coils added up to 28. The room available prohibited adequate spacingbetween the coil and shield can. This plus the need for all the availabletuning range dictated the use of powdered iron sleeves as magnetic shields.These sleeves then permitted the electrostatic shield can to be placeddirectly next to the sleeve without serious effect on the coils.

The development of powdered iron cores for this frequency rangerepresents a project in itself. It was not difficult to obtain cores pos-sessing sufficient permeability and loss characteristics above 100 kc,but the lower frequencies presented what looked like an insurmountableproblem. The iron core companies approached on the problem had given upon the lower frequencies. Powdered iron did not exist with losses highenough to lower Q as the core was inserted into the coil.

Experiments at Collins met with various degrees of success. A build-up of transformer iron laminations possessed sufficient permeability butthe losses were too great. The Q factor would all but disappear at about1/4 full insertion. A composite core was made by gluing transformer lami-nations of various lengths to pieces of powdered iron cores. It was foundthat by selecting the lengths of powdered iron and laminations that a widevariety of Q effects could be obtained. By using various laminationmaterials - Audio "A", oriented silicon, etc. - the effect was further con-trolled. Unfortunately time was not available to develop means of pro-ducing a core of this type in quantity.

Abandoning the composite iron core, the lowest frequency coresavailable (100 kc) were modified by wrapping copper bands around them atvarious intervals. By varying the position of the bands, the number used,and their width, it was possible to control the Q. Here again the pro-duction possibilities appeared negative. A solution was secured by coating

45


RESTRICTED

the cores with bands of a silver conductive coating. Dupont 4817 was used,and a typical coil and core can be seen in Figure 15. This does not pro-duce an absolute linear curve, but is sufficiently close for this appli-cation. This core can be produced in quantity.

A remote control pentode was selected for the RF amplifiers. Thereasonably quiet 12BA6 was used in the first development set and changed tothe nearly equivalent 6BJ6 as a result of heat considerations. The balancedantenna input was specified as 125 ohms. Previous mention has been made ofthe QX being partially determined by sensitivity requirements, The antennagain is approximately 5. A calculation was made for each end of each bandand for each unbalanced antenna impedance combination. This showed thatonly at the highest frequencies, with the highest antenna impedances, andhighest capacitive antennas could over coupling occur if the antenna wasconnected directly to the top of the coil. Highest gain resulted fromthis connection, being as high as 224 at 28 kc but showing a net of 2.53due to mismatch. At 1500 kc with an antenna of 1000 micro-microfarads and0.1 ohm the net gain is 64.3. This antenna would be very rare. Had over-coupling been possible in many cases, it would have been necessary to usea different series matching capacitor on each band.

In order that the antenna impedance remain constant over each band,it was necessary to provide a link of unusual construction. On the twolowest bands the link winding is wound directly on top of the main windingand is a pie type coil. The impedance is held constant by providing morethan one pie at strategic locations and of an appropriate number of turns.The design of such a device is done by experimental methods only. Thehigher band links are solenoids wound on bakelite forms which in turn slipover the main coil winding. Here again the placement of turns is variedto hold the impedance constant. Electrostatic shielding is obtained bycoating the inside of the link form with a silver conductive paint andgrounding the shield thus formed. There is of course an open strip in thiscoating.

The specified balance ratio of 500 to 1 was secured by connectinga fixed capacitor from one end of the antenna link to ground. Thiscapacitor is not needed on all bands.

The tracking of the RF coils is a function of both electrical andmechanical tolerances. The design was such that the overall trackingerror should be held to less than 1.5 db. This is at room temperature andnormal climatic conditions. It is felt that in production variable pitchprogressive universal windings will improve this error or at least makeit easier to maintain.

The two coils in the plate circuit of the second RF amplifier weretop coupled by capacity. In order that the gain of the coupled systemvary as little as possible over the band, the coupling was set at 0.66 atthe low frequency end. This gave a coupling factor of 2.0 at the highfrequency end and the gain variation was 1.0 db. The over-coupled con-dition of these coils was found to adversely affect the direction findingbandpass phase shift characteristics.

46


RESTRICTED

During the tests made at Coles Laboratory on the first receiver, itwas found that the cross-modulation characteristics of the receiver couldstand improvement. The effect was particularly bad when using an un-balanced antenna which essentially swamped the antenna coil. As a resultit was decided to place two coils in front of the first RF grid. Thiseliminated the coupled circuit in the plate circuit of the second RFamplifier; and, of course, opened up the possibility for more harmonicdistortion from this tube reaching the first mixer grid.

It was found that if the two coils were placed in the first RF gridcircuit and top coupled by capacity, the antenna input impedance fluctuatedwildly. A very small tracking error in either coil would throw a largereactive component into the antenna circuit. Final consideration re-placed the capacity top coupling with resistive top coupling. With sucha circuit over-coupling is impossible. Unfortunately there is a voltageloss and sensitivity suffered accordingly. The 10 db, 30% modulated,signal-to-noise ratio dropped from 3 microvolts to 6 microvolts at an 8 kcbandwidth. Tests made with the Signal Corps showed the improvement incross-modulation worth the loss in sensitivity.

It was necessary to devise some system for switching the RF coils.In the low frequency range - 15 kc to 500 kc - five RF bands are used. Inthe high frequency range - 500 kc to 1500 kc - two RF bands are used. Justhow to switch these bands at the right time presented a major problem. Amechanical system to do this would be a monstrocity since variable overlapwould have to be provided. An electromechanical device was finally in-corporated to perform this function. Reference to the mechanical driveshown in Figure 71 will provide full information. Briefly, the operationis as follows:

1. Rotary information from the oscillator shaft is supplied to an electrical commutator coupled thereto. On this commutator the gaps represent overlap and the segments RF bands. Two commutators are used; one for each range.

2. The commutator is connected to a selector switch section in the bandswitch assembly. This switch controls a motor which rotates the bandswitch when energized.

3. The motor stops when the proper band is switched into the circuit.

4. The operator selects one of two ranges - 15 kc to 500 kc or 500 kc to 1500 kc - the appropriate commutator being switched into the circuit for each range.

The second RF amplifier plate coil is similar to the antenna coilin its link construction. A different turns and winding data is used andthere is no static shield.

47


RESTRICTED

A triode first mixer was chosen because of the superior noiseresistance characteristics. This determined the necessary gain of theRF amplifier stages. An approximate gain of 5 from the first RF grid tothe mixer grid was needed. Since the 6BJ6 and a 5000 plate load can pro-duce a gain of about 20, the output was resistance divided before beingcoupled to the second grid. The net gain - grid to grid - was then 3·5.·The gain of the second RF stage grid to the mixer grid was centered at 1.5.The first RF gain was made higher than the second for noise considerations.

The second mixer is a pentagrid (6BE6) which has a high noise re-sistance. There is no IF stage between the two mixers; only a coupledcircuit. This transformer is approximately 100 kc broad at the nose with0.6 coupling, giving a stage gain for the first mixer of approximately 2.

The second mixer plate load is tuned and has an untuned balancedsecondary. This secondary drives the crystal filter input circuit of the455 kc IF strip. Coax cables are used to couple the two units.

Temperature runs were made on all of the coils in the signal path.The drift was low enough to not warrant temperature compensation.

Incorporated in the RF unit is a "Test Oscillator". It is possiblewith this oscillator to align the RF coils without the aid of a signalgenerator. The oscillator is a transitron providing two terminal oper-ation. Thus it is possible to connect an RF coil to the oscillator with-out the need of a feedback winding. By using the BFO and the tuning dialthe RF coils can be resonated at prescribed frequencies for alignment.

It was decided to use this oscillator for two reasons:

1. Signal generators are notoriously inaccurately calibrated in the very low frequency ranges. The manufacturers generally strive to provide accuracy in the more widely used ranges.

2. The very low Q coils used in the R-389 make accurate alignment tedious if not impossible. A small error in each coil may seriously impair the band pass of the front end.

Unfortunately this oscillator does not oscillate at the exact fre-quency of anti-resonance. This error is small, however, and consistent;and it is felt that the service advantages provided by the oscillatoroutweigh this disadvantage.

The Injection Path

As stated before the frequency accuracy and stability of the 10,455kc crystal oscillator is unimportant because of the ring canceling action.A complete block diagram of the injection path is shown in Figure 2. It

48


RESTRICTED

will be noted that the oscillator feeds injection to the second mixer andpush-pull injection to the injection mixers. This is accomplished by meansof a transformer, the secondary being balanced. The injection mixers arebalanced in such a manner as to cancel the 10,455 kc input. This is neces-sary due to the possibility of a spurious response at 455 kc if the 10,455kc is permitted to reach the first mixer. The injection mixers are fedbalanced input at the VFO fundamental or second harmonic from a phase in-verter. The phase inverter is driven by the VFO buffer from a tuned plateload which is tuned either to the fundamental or second harmonic.

The injection mixer plate load is tuned and the two frequency rangesare covered by two coils. This plate load is coupled to a cathode followeramplifier called the input coupler. This input coupler drives an "m"derived, low pass filter at its characteristic impedance. The filter isdesigned to have maximum attenuation at 10,455 kc (50 db) and cutoff fre-quency is 10,000 kc. This is to further reduce the level of 10,455 kcreaching the first mixer, but must pass the injection range of 8495 kc to9985 kc.

The filter output is coupled to a grounded grid amplifier called theoutput coupler. The output coupler plate circuit is tuned to the in-jection frequency as is the plate circuit of the injection mixers.

The output coupler plate circuit is coupled to a cathode followercalled the mixer driver. The mixer driver has an untuned transformer inthe cathode circuit and is used to drive the cathodes of the first mixer.

The first mixer is balanced. The balance circuit is such that theinjection is canceled in the plate circuit. This is necessary to preservethe signal-to-noise ratio below 100 kc and is adjustable by means of acathode balance potentiometer in the first mixer. As the input frequencylowers below 100 kc, the injection frequency approaches the 10 mc firstIF frequency; the result of which is transmission of noise from the in-jection path into the signal path. Cancellation of the injection voltageminimizes this effect.

The 28-Volt power Supply

A dry disc rectifier, power transformer, and capacity filter comprisethe 28-volt power supply. This supply furnishes power to the band-switching motor. The commutator circuit causes a tube to conduct whichin turn energizes the relay controlling the power supply.

4·2.5.3 Mechanical Considerations

A complete report of the mechanical design is beyond the scope orintent of this report. However, several major points will be discussed.

49


RESTRICTED

The Oscillator Stop

Since the receiver was designed to cover its entire spectrum in twosweeps, it was necessary to provide considerable vernier action within theoscillator proper. Approximately 50 turns of the oscillator are requiredto cover its range. The usual tab washer stop would be too long for thismany turns. A unique gear and pawl system is used to provide a secure stop.Further gear reduction of the oscillator gives a total of 208 turns requiredat the tuning knob to cover the range. This requires a stop of relativelyhigh strength since the stop is located on the oscillator shaft. Thisstop is shown in Figure 45.

The Signal Corps was of the opinion that 208 turns required at thetuning knob was excessive for search operation or quick frequency change.Accordingly, we incorporated a motor tuning system to assist the operator.This is a clutch operated reversible system and will tune a complete rangein less than one minute. This can be seen in Figure 13.

A cam driven slug rack system providing a travel of 1-1/4 inches wouldrequire excessive space. Reversible lead screws similar to those used infishing reels, were used in place of cams. These proved to provide satis-factory linearity without special tooling for their manufacture. The leadscrews are gear driven from a line shaft system. This technique has theadvantage of operation in parallel rather than series and so does notbuild up large amounts of backlash.

Construction techniques were governed by the environmental require-ments of the specification. Vibration and shake tests caused additionalstrengthening to be added in later models.

4.2.6 R-390 RF Circuits

The approximate size and shape Of the R-390 RF unit had been determinedby the early planning of the main frame and sub-units. The general fre-quency scheme had also been selected and the work of detail designing couldbe started on such things as the coils, crystal oscillators, and mixers.This work, along with the mechanical drives in these units, was the mostcomplicated and took the longest time of anything in the project. Thefinal RF unit is shown in Figures 35, 36 and 36A.

4.2.6.1 General Design

The RF unit was designed to provide frequency coverage from .5 - 32mc with good sensitivity, a minimum number of bad spurious responses, andhigh rejection to image and IF frequencies. The circuit is a triple super-heterodyne below 8 mc, and double superheterodyne above 8 mc. This typeof circuit was chosen for two basic reasons. In order to provide goodstability and dial accuracy, the variable frequency oscillator must be aprecision device, operating at a fairly low frequency and with a limitedrange (1 mc). Use of such a VFO requires the use of an extra conversion

50


RESTRICTED

stage to convert incoming signals to a frequency range which can be usedin conjunction with the VFO to produce a 455 kc IF. In order to reduceimage response to a satisfactory level at the highest ranges of the re-ceiver this variable IF frequency should be between 2 and 3 mc. Usingthis type of dual conversion, some bad 3rd order spurious responses willbe encountered between 4 and 6 mc, and 1st order spurious responses be-tween 2 and 3 mc. This makes it necessary to go to triple conversionbelow 7 mc in order to avoid these bad spurious responses. By means oftriple conversion it is possible to reduce all spurious responses to 5thorder or higher.

Space limitations on the size of the RF chassis made it necessary tolimit the number of RF coil ranges to six. Because of the multitude oftuned circuits, slug tuning is the most practical way of varying the fre-quency, since as many as 10 variable tuned circuits are in use at thesame time. Figure 2 is a block diagram illustrating the operation of theRF unit.

4.2.6.2 Coil Design

Applicable Specification

Specifications applying most directly to coil design are:

a. RF circuit bandwidth

b. IF and image rejection ratio

c. Service test effects on sensitivity

d. Sensitivity

e. Antenna input circuit

Design Consideration

It has already been mentioned that six RF coil ranges is the maximumnumber that could be accommodated on the RF chassis. As it turned outthis is also the minimum number that could do the job. Using high gradeiron cores, which are little affected by climatic changes, it is difficultto get more then 2-1 frequency coverage in any coil, particularly when theslug travel is also limited, as it is in the R-390 receiver. Figures 39and 40 show constructional details of the RF coils.

Slug Tuning

Slug tuning has a number of theoretical advantages over capacitivetuning among which are these. By proper choice of iron, the bandwidth ofany coil may be made nearly constant throughout its range. This conditionalso produces a constant impedance coil. For a frequency coverage of 2 to 1,

51


RESTRICTED

the core should be so chosen as to reduce the Q 2 to 1 as the slug variesfrom its position at the highest frequency to its position at the lowestfrequency. Then BW1=f1/Q1 represents the bandwidth at f1, the low fre-

quency end of the coil. The bandwidth at f2, the high frequency end, is

f2 2f1BW2 = = = BW1 Q2 2Q1

If Q varied exactly as the inverse of f, the bandwidth would be con-stant throughout each coil range. This condition is only approximated inpractice since the Q drops relatively too fast with frequency at first.

If Q varied exactly as the inverse of f, coil impedance would also beconstant throughout the frequency range since Z1 = Q1X1 where the subscript

indicates the low frequency end of the coil range. Then Z2 = Q2 X2 but

Q2 = 2Q1 and X2 = X/2

2Q1X1so: Z2 = = Z1 2

Again, since this condition is only approximated, the gain of the re-ceiver is not absolutely constant throughout each coil range, but ingeneral, varies between 1-1/2 and 2-1/2 to 1, depending on the coil rangein use.

Number of RF Coils

The number of RF coils to be used in any one band was set at four,including the antenna coil. It was thought that this would provide ade-quate protection against IF and image frequencies. As it turned out itwas impossible to get the required Q in the 16-32 band and consequentlythe image response occasionally drops below 80 db around 24 mc, where thebandwidth is the widest.

Type of Winding

Originally the coils were wound at a constant pitch, and the camsdriving the slug rack were cut to give a constant frequency change withrespect to angular rotation. One receiver was built using this principle,but the scheme was abondoned because in order to work right the mechani-cal position of the cam had to be perfect. If it were off ever so slightlythe coil would not track. Electrical alignment of the coil is also verydifficut with this scheme, and no simple alignment procedure could be found.Linear wound coils may have been practical if less coverage were requiredor more slug travel available.

52


RESTRICTED

All receivers since the first model use variable pitch coils in con-junction with linear rise came. Because the cam has linear rise anywhereon its usable portion, exact cam positioning is not critical. Variablepitch coils also allow a simple alignment procedure. Coil inductance isadjusted near the low frequency end of the coil range, while coil capaci-tance is adjusted near the high end of the range.

Up to 4 mc it is necessary to use a multilayer coil to obtain therequired inductance. These coils were originally designed using ninepi's equally spaced with different number of turns in each pi. This typeof winding has two disadvantages. Because the coil consists of lumpedsections, the tracking cannot be made absolutely linear, but instead thetracking curve is a wavy line varying plus or minus about the desiredstraight line. The second disadvantage lies in the difficulty of massproducing this type of winding. Each pi must be wound separately andspacing between pi's must be critically maintained.

Variable pitch progressive universal winding overcomes both ofthese difficulties. Because the coil is continuous instead of being asum of discreet sections, tracking is improved, and the coil can be en-tirely machine wound. All future multilayer coils should be designedusing VPPU windings.

Coil Q

Coil Q was largely determined by the requirement of RF circuitbandwidth. Through the broadcast band where strong signals abound, thisconsideration was made secondary to cross-modulation protection, The RFbandwidth at 550 kc is 9 kc to the 3 db points and at 1.100 mc it is 10kc to the 1 db points.

Above the broadcast band, RF selectivity adds only a fraction of adb attenuation within the IF passband, assuming the coils are tuned upproperly.

In an attempt to raise the Q of the 16-32 band coil as high aspossible, the RF coils were tapped, reducing the tube loading on thecoil. Tapping also has other beneficial effects such as reducing strayinductance and capacitance which reduces the tuning range of the coil.In addition, lowering the load impedance in the RF tube plate circuitserves to improve stability which becomes more of a problem at highfrequencies.

In general, the Q at the high end of the coil range is largelya function of wire size and design of the variable pitch winding. Thesehave been optimized in the 16-32 mc band to provide the highest possibleQ. At the low frequency end of the coil range the Q is primarily de-termined by the characteristics of the iron core. Much time was spenttrying to find the best slug for a given frequency range. The threeimportant factors in the choice of core material were Q, permeability

53


RESTRICTED

and stability under climatic variations. Finally the core materials werereduced to 3 types manufactured by Stockpole Carbon Co.; namely, S-100A,S-62, and S-51. The S-100A is used in the 8-16 and 16-32 mc RF coils andin the 9-18 mc VIF. S-62 iron is used in the 3-2 VIF and 4-8 mc RF coils.S-51 iron is used in the .5-1, 1-2 and 2-4 mc RF coils.

Below is a table showing coil Q's at the high and low ends of eachcoil range.

TABLE VIII

RF Coils VIF Coils f Q f Q f Q_

.5 33 4 44 2 75 1.0 54 8 74 3 75 1.0 28 8 53 9 53 2.0 54 10 98 18 98 2.0 25 16 47 4.0 54 32 108

Q tolerance can be about ±±10%. Effective operating Qs in the set aresomewhat lower especially at the higher frequencies. The coil Q of the 9-18mc IF coincides with the 8-16 mc RF coils, since they are identical. The Qof the 3-2 mc variable I-F was chosen to give high attenuation to the secondcrystal oscillator frequency while maintaining adequate bandwidths.

Coil Impedance

A nominal value of 20,000 ohms was chosen as a design figure for theRF coils. This gives reasonable values of inductance and capacitance through-out the whole frequency range of the receiver.

Tracking Tolerance

Electrical requirements on the RF coils necessitates a tracking toler-ance within .4 db per coil. Additional mechanical tolerances raise this to atotal of 1 db per coil. The coupled variable I-F stages broaden the tuningof these stages so that only a 1 db tolerance is necessary for the sum ofall 3 VIF coils. Assuming any mistracking in the antenna coil is tuned outby the antenna trimmer, the total mistracking allowance is 3 db for the RFcoils plus 1 db each for both of the variable IFs, or 5 db maximum.

L.2.6.3 Antenna Input Circuit

Taking up first the consideration of sensitivity it is evident that thehigher the antenna coil gain, the higher will be the receiver sensitivity.If sensitivity were the only consideration, it would be desirable to makethe antenna coil gain as high as possible. Other problems such as receiver

54


RESTRICTED

overload at high input levels and cross-modulation make it necessary tocompromise. Then, also, at the high frequency range (16-32 mc) the im-pedance of the antenna coil is limited both by the inevitable stray capaci-tances and the RF losses of switches, tube sockets and input impedance ofthe first RF tube. Because of these stray losses and stray capacitances,the impedance across the antenna coil is on the order of 10,000 ohms above16 mc and approximately 20,000 ohms below this frequency. With an inputimpedance of 125 ohms, this represents a coil gain of 4.5 and 6.3respectively. Because of mismatching, these figures are seldom attained.

Antenna Coil Primary

In order to provide electrostatic shielding between primary andsecondary, the primary is wound inside split brass rings mounted coaxiallywith the coil. Figure 39 shows the construction of the antenna coil.It was found by experiment that if all the turns were placed in a singlering, no position of the ring could be found which gave anywhere near con-stant impedance over the tuning range. Splitting up the winding into tworings suitably spaced provided a reasonable compromise between simplicityof construction and constancy of antenna input impedance. However, insteadof 125 ohm ±±15% the impedance actually varies from 50 - 300 ohms. Thewidest variation occurs within the 16 - 32 mc band antenna coil. There areonly three primary turns on this coil and positioning of the rings is quitecritical. Despite the use of electrostatic shielding between primary andsecondary, careful shielding underneath the RF chassis between primary andsecondary coil terminals along with an antenna balancing trimmer, it hasnot been possible to obtain the 54 db balance to unbalance ratio required.The following table shows approximate balance to unbalanced ratios obtained:

__Band__ Balance Ratio

.5 - 1 45 db

1 - 4 40 db

4 - 8 35 db

8 - 16 30 db

16 - 32 25 db

Unbalanced Antenna Input

According to the specification, the unbalanced antenna input is to beused both for single ended whip antennas and doublet antennas using 50 ohmcable. In order to accomplish matching from a short whip to the antennacoil at low frequencies, loading coils are necessary to tune out the capaci-tive reactance of the whip. In order to be effective these loading coilsmust have low losses, and tuning becomes quite critical. Besides these

55


RESTRICTED

problems the compactness of the RF unit left no room for such coils. Becauseof these factors it was decided not to attempt to match perfectly for whipantennas at low frequencies.

The other consideration in designing the unbalanced input system wasthe problem of providing reasonably good matching for low impedance re-sistive antennas (half-wave doublet, folded dipole, etc.). Over the range1 - 16 mc this has been accomplished fairly well, with resistive inputsvarying from 42 - 360 ohms in this region.

Both types of unbalanced input can be accommodated reasonably well withcapacitor coupling. An antenna trimmer has been provided to tune out re-actance reflected across the coil from either the balanced or the unbalancedinput.

In summary, these are the considerations that governed the choice ofunbalanced input coupling capacitors:

(a) Coupling capacitor shall be limited to such size as will permitcomplete trimming out of reflected reactance of any capacitive antenna.

(b) Within the limits set up above, the capacitors shall be chosento provide a good match between the whip antenna and the antenna coil.

(c) These values of capacity should be modified slightly if necessaryto provide good matching for low impedance resistive antennas.

(d) Two important exceptions are noted. These are:

(dl) Consideration (a) would lead to a capacitor in the .S - 1 mc band considerably too small for consideration (b) and (c). Therefore, it was increased 33 per cent at a sacrifice of complete tuning out of resistance for low impedance capacitive or resistive antenna. Since such antennas are not often encountered in this frequency range, such a situation is not serious.

(d2) Stray capacitance limits the size of the 16-32 mc antenna coupling capacitor to 5 uuf. This value is too small to provide adequate coupling, and consequently sensitivity suffers slightly in this range.

Static Charge Drain Circuit

In accordance with the specification a static discharge circuit wasincorporated consisting of a 220 k ohm resistor in parallel with a neon lampbetween the unbalanced antenna terminal and ground. The resistor drainsoff charges which don't have enough voltage to fire the neon lamp, whilethe latter provides a rapid discharge for a suddenly applied high voltage.

56


RESTRICTED

Circuit Diagram

A functional circuit diagram illustrating the antenna input circuitis shown on Figure 63. Note the capacitor in series with the ant trimmercapacitor. This is necessary with three of the coils to reduce the totalcapacity across the coil. Except on the 16-32 mc band it does not seriouslyaffect the ant trimmer effectiveness. Although not shown on this simplecircuit diagram, the coil preceding the one in use is shorted out toeliminate "suck outs" which would otherwise occur.

4.2.6.4 RF Stages

From the specifications, the three most important considerations in thedesign of the RF stages are noise contribution, gain, and strong signalhandling capabilities.

Noise Contribution of RF Stages

Once the gain of the antenna coil has been fixed, the next method ofraising the sensitivity of the receiver is to use a low noise first RF tube.All low noise pentodes, however, are sharp cut-off tubes, and applicationof full AGC to these tubes results in severe distortion above about .1 voltantenna input. A one volt signal 30% modulated at the antenna has a possiblepeak voltage of 11.5 at the grid of the first RF tube with the antenna coilsused. The first RF tube must be able to handle this voltage without seriousdistortion, and in order to prevent the grid from going positive with sucha voltage, at least 11.5 V of bias must be supplied to the tube. For mostsharp cut-off tubes, this lies far beyond cut-off, but the 6AJ5, thoughnominally a short cut-off tube can be operated so that it has fairlysatisfactory remote cut-off characteristics. This is primarily due tothe fact that it was designed for 28 V plate and screen operation. Ap-plication of screen voltage to this tube through a large screen droppingresistor provides proper operation when used with a 180 V B+ supply. WhenAGC is applied to the control grid, the screen voltage rises, thus pre-venting an early cut-off and giving a much better remote characteristicthan any other sharp cut-off tube tried.

As an example of its superiority, here is a comparison between it andthe 6AK5:

Distortion at .7 V input (30% mod)

6AK5 16% 6AJ5 3%

Audio rise from 10 uv to .7 V (db)

6AK5 18.5 db 6AJ5 10.0 db

57


RESTRICTED

Despite the use of this tube, sensitivity at times rises above3 uv, due to variations in tubes, ant coil gain, etc. It is proposed, there-fore, to raise allowable AM sensitivity to 3.3 uv below 14 mc and 4.4 uvabove 14 mc. Allowable CW sensitivity is to be raised to 1.3 uv above 14 mc.

The cascode amplifier was also tried on breadboard models and seemedto offer some improvement over pentodes in regard to sensitivity, but again,their poor performance when AGC is applied disqualifies them. They areoften used in TV sets with AGC applied, but they will not handle signallevels of 12 V peak as required in this receiver.

Gain

Closely allied with a low noise first RF tube in the matter of noisecontribution is the gain provided in the first RF stages. For greatestsensitivity the first RF stage should have a high gain, but such high gainis undesirable from other standpoints. The signal handling capability ofthe second RF tube does not greatly exceed that of the first RF tube. Thismeans that the gain of the first RF tube should be limited to unity when a1 V signal is applied to the antenna. If the AGC voltage reduces the firstRF tube gain by 10 with 1 V of antenna input, then the maximum upper limitof gain below AGC threshold is also 10. In order to hold down cross-modu-lation and reduce the magnitude of spurious responses, low gain is alsodesirable. From these considerations a nominal gain of five was chosenfor the first RF stage. With this value of gain, the noise contributionof the second RF stage is not very great and it becomes possible to choosethe tube type on the basis of other considerations. These considerationsare distortion at high inputs, freedom from cross-modulation and good AGCcharacteristics. A remote cut-off pentode is desirable from all thesestandpoints and consequently a 6BJ6 was chosen for the second RF stage.

It was stated above that spurious response was affected by gain. Bestsignal-to-spurious ratio is obtained if signal level at mixer grids is heldlow. On the other hand best sensitivity results with high RF gains. Acompromise was reached at a gain of 25 from antenna to first mixer grid.This allows a gain of approximately one in the second RF stage. With thesame overall gain of 25 to the first mixer grid, dividing the gain in theRF stages into five for the first RF and one for the second gives bettersensitivity than any combination using a gain larger than unity in thesecond RF stage. Considering the noise produced at the mixer grid by thefirst and second RF stages, a constant gain from first RF grid to firstmixer grid results in a fixed noise contribution from the first RF tubeirregardless of the division of gain between first and second RF tubes.The noise contribution of the second RF stage to the total noise at themixer grid is the second RF stage noise multiplied by the gain betweensecond RF grid and first mixer grid. Obviously the smaller this gain theless noise contribution by the second RF stage.

58


RESTRICTED

Strong Signal Handling Capability

Some of the items planned for discussion under this heading havealready been taken up. These include the superiority of the 6AJ5 overother sharp cut-off tubes with respect to AGC control and strong signaldistortion, and the choice of the 6BJ6 as the second RF amplifier fromthe standpoint of better AGC characteristic, lower distortion, and bettercross-modulation characteristics. It would be well to note here that the6AJ5 is also very good in regard to cross-modulation for interferingsignals up to .1 volt. At higher signal levels it deteriorates below theperformance of remote cut-off tubes, particularly at levels around 1 volt.

It may be noted that AGC is supplied to the RF tube grids throughhigh values of resistors. These help to prevent blocking of the receiverby strong signals outside the IF pass band, but within the RF pass band.These strong off-tune signals may cause the first or second RF tubes todraw grid current which will back up into the AGC line. By placing highresistors in series with the grids, little of the grid leak bias so de-veloped is fed back to the IF controlled tubes.

Tests on the strong signal handling capability of the receiver revealthat with AGC on and RF gain wide open, the blocking level always exceeds1 volt.

4·2.6.5 Mixers

General Considerations

The over-riding consideration in the choice of mixer tubes was re-jection to spurious responses. The problem resolved itself simply intothis: with RF gain adjusted as necessary for a given sensitivity whichmixer tube gives the best rejection to spurious responses? Considerableexperimentation was done on mixers from which the following conclusionswere drawn: All mixers are very level conscious in regard to productionof spurious responses. In order to get measurements that can be dupli-cated, the reference input was standardized at 5 uv by agreement with theSignal Corps. For the same input level at the grid of the mixer tube,pentagrid mixers are superior to triodes and pentodes. However, the re-quirement for the same signal-to-noise ratio irregardless of the mixertype required that greater RF gain be used in conjunction with pentagridconverters, When the RF gain was adjusted for each tube type to give thesame signal-to-noise ratio, triodes actually gave the best rejection tospurious responses. Among different types of triodes, there are variations,and the 6C4 was found to be the best of all tubes tried. This is based onan average since there is considerable variation among tubes of the sametype. For every tube tried, bias and oscillator drive were adjusted foroptimum spurious rejection. Careful adjustment of the bias and/or oscil-lator drive could reduce a spurious down 20 db or more below normal level.Another tube of the same type would require a slight readjustment to ob-tain the same results. Then too, this particular adjustment of the bias

59


RESTRICTED

would have no effect on a spurious of different order which would requirean entirely new readjustment. The most that could be done was to find abias, which, aside from these unusual "holes" gave the best spurious re-jection for all orders of spurious responses likely to give trouble.

Many different schemes of balanced mixers yielding useful outputwere tried, but they offered no significant improvement over single endedtypes. It is possible that this may have been largely due to unbalancebetween the two mixer tube sections, and also to the difficulty of obtainingcircuit balance at the frequencies involved.

Oscillator Coupling

Many methods of coupling oscillator and signal voltage into the mixerexist, but the following are the considerations which led to cathode in-jection:

One side of the signal source is normally grounded. Unless parallelgrid injection was used, the oscillators would have to be fed in serieswith the signal on the grid side. Parallel injection involves a loss ofsignal or oscillator voltage, or both. Series injection on the grid sideinvolves keeping both oscillator leads above ground, and with littlecapacitance to ground in order not to detune the signal coil.

Coupling the oscillator into the cathode removes the above objections.There is some coupling of the oscillator voltage to the grid due to gridcathode capacitance with this scheme, but it never exceeds 15 per cent andcauses no difficulty.

High Variable IF

The high variable IF frequency range was determined by several con-siderations.

Oscillator Crossovers

In order to eliminate crossovers of the type 1nX1 - mX21 = VIF, thehighest common factor of X1 x X2 must be no lower than four except in the.5 - 1 mc band where it may be three. In this equation X1 and X2 are re-spectively the frequencies of the first and second crystal oscillators. Mand n are any two integers required to satisfy the equation. The VIF of theequation is the 3-2 VIF. If X1 and X2 have a highest common factor no lessthan four, it is impossible to find any integers m and n which will satisfythis equation, and there will be no spurious signals generated by harmonicsof one oscillator beating with harmonics Of the other oscillator. In the.5 - 1 mc band the VIF varies from 2.5 - 2 mc. This allows the highestcommon factor to drop down to 3, which with the 3 - 2 VIF also is thehighest possible common factor.

60


RESTRICTED

An exception was made on band 3 in order to sidestep a 4th orderspurious. X1 and X2 were chosen so that n = 10 and m = 20 before adifference of their harmonics lies in the 3 - 2 VIF. These are highenough so that no trouble is caused.

Spurious Crossovers

The high IF should be chosen so that no spurious crossover lowerthan 5th order exists.

Frequency Range

All high IF frequencies must be within a two to one frequency rangeto permit tuning with a single coil.

Number of Coils

Although the frequency scheme was chosen to eliminate whistlescaused by crystal oscillator harmonics, the crystal oscillators can, ifthey get into the wrong mixers, cause spurious responses. Three high VIFcoils were chosen as the minimum number required to provide adequateprotection of the second mixer from first crystal oscillator voltage fedinto it via the high VIF. Three coils provided a mixer gain of slightlygreater than one which is adequate.

3-2 mc VIF

The 3-2 VIF design was determined by the following considerations:

Image Rejection - The frequency should be high enough to provideadequate image rejection in the highest frequency band.

Stability - in order to get good receiver stability the VFO mustoperate at a fairly low frequency. The VIF must then be 455 kc awayfrom this VFO frequency.

Number of Coils - Again three coils were chosen both to provideadequate attenuation to the second crystal oscillator frequency and toprovide additional selectivity for the last mixer at high frequencies.Despite the high Q of these coils, they contribute little attenuationwithin the passband of the 16 kc IF bandwidth position. The circuitwas designed to approximate the maximally flat case. The bandwidth tothe -.5 db points is 27 kc at 2.1 mc and about 50 kc at 3 mc. Gain inthis mixer stage is normally 2.

455 KC IF

The 455 kc IF coil in the plate of the last mixer was designed togive the maximum possible gain consistant with the desired bandwidthand output impedance. Gain of the last mixer stage is nominally 1.7.

61


RESTRICTED

Spurious Rejection

Despite all the experimentation undertaken on mixers and frequencyschemes, no means was found to reduce spurious responses to the low leveldemanded by the specification. However, all spurious responses have beenreduced to 55 db or better.

4.2.6.6 Miscellaneous Considerations

Parasitic Suppressors

Under certain conditions, parasitic oscillations have been discoveredin the second RF tube, first, and second mixers. These were caused mainlyby long grid and plate leads, the tube acting as a tuned-plate tuned-gridoscillator at frequencies in the hundreds of megacycles. Placing a 27 ohmresistor in series with the grids lowered the Q of the grid line low enoughto prevent oscillations.

Test Points

As an aid in troubleshooting and to facilitate gain measurements inthe RF, test points are provided leading to each tube's control grid. Thisis a decided help since the underside of the RF unit is not readilyaccessible.

Lead Filtering

Each RF and mixer stage is decoupled from B+ by means of a 2200 ohmresistor and an .01 uf capacitor, except for the first RF tube which hadits plate decoupled with a 1 uf capacitor. The purpose of the capacitor isto prevent audio frequencies on the B+ line from remodulating the RF signal.Initially when the set was tuned to the side of a strong carrier, it brokeinto an audio oscillation. A small disturbance on the B+ line would modu-late the B+ line creating sidebands within the passband of the receiver.Detection of the composite signal feeds audio voltage back to the B+ linewhere it remodulates the RF tube and maintains the audio oscillation. Allmixer tube filaments are thoroughly filtered to prevent the oscillatorsfrom getting into wrong mixers and causing spurious responses.

RF Gain Control

Manual RF gain control is obtained by varying the cathod resistorsof the RF amplifiers and two of the IF amplifiers. The RF gain control isbypassed with a 50 uf electrolytic capacitor to reduce noise induced byrotation of the control.

4·2.6.7 The Crystal Oscillator Unit

The crystal oscillator unit is shown in Figures 41 and 42. This unitcould have been part of the RF unit but was made a separate unit to alloweasier manufacture and maintenance.

61


RESTRICTED

General Design

The basic oscillator circuit used is the pentode cathode-feedbackPierce oscillator. Since the screen is grounded, the plate is fairlywell shielded from the input and tuning the plate circuit has negligibleeffect on oscillator frequency. The plate tank is tuned to either thefundamental, second harmonic or third harmonic of the crystal as the RFunit requires. This is necessary since JAN specs allow no CIi-36U crystalshigher than 15 mc while the RF unit requires frequencies up to 34 mc.Overtone crystal oscillators using the series resonant frequency of thecrystal were not considered because of their poor stability and low outputvoltage.

Figure 64 shows the schematic of the second crystal oscillator inthe band 34 switch position. Although there are 32 bands, only 18 crystalsare used in the second oscillator since some are used more than once.Also there are only 25 output frequencies tuned with 24 trimmers sincebands 0 through 7, with the exception of band 3, use frequencies which areused on some of the other 24 bands. Band 3 requires an output frequencyof 18.6 mc, which is obtained by using the tuned circuit for 18 mc as usedon bands 6 and 15 with a series capacitor switched in on band 3 to raisethe resonant frequency to 18.6 mc.

The first crystal oscillator, operating only on bends 0 to 7, usesfive crystals, three of which are used twice, and five trimmers to tunethem. This oscillator operates only on fundamental frequencies of thecrystals.

Choice of Tubes

After experimentation with a number of different pentodes, the 6AJ5was found to give the required output voltage with the least power drain.

Plate transformers

In the plate of each oscillator tube, is a transformer for peakingthe output at the desired frequency and supplying voltage at low impedanceto the mixer cathodes. Because of the very wide range it covers and be-cause at various times it is tuned to fundamental second or third harmonicof the crystal, the output of the second oscillator varies 3 to 1 involtage over its frequency range. Since the second mixer is fairly wellsaturated, the receiver gain varies only about half this amount. Becauseof its much narrower frequency range the first crystal oscillator outputvoltage varies about half this amount. One of these transformers isshown in Figure 65.

63


RESTRICTED

Oscillator Ovens

Measurements indicate that the chassis of the R-390 receiver risesabout 20 deg C above the ambient temperature under continuous operation.Since the receiver is to operate in ambient temperatures up to 65 deg C,an oven temperature of 85 deg C is necessary to insure that oven controlis effective up to this temperature. CR-36/U crystals designed for thisoperating temperature have been used for the crystal oscillator. The con-struction of the oven can be seen in Figure 42. It is made up of an innerand outer metal box with a 50 watt heating element wound directly on theinner one. An 85 deg C thermostat is also mounted directly on theinner box with about 3/8 inch of fiberglass insulation between boxes. Theinner box is made of heavy brass to give good head distribution and ther-mal inertia. With this oven the crystal temperatures are held to lessthan two degrees temperature variation.

Lead Filtering

A number of spurious responses in the receiver are caused by oscil-lators getting into the wrong mixers. In order to reduce leakage fromone crystal oscillator to another, careful shielding is employed betweenoscillator sections and common B+ and filament leads are well filtered.

4.2.7 R-389 Variable Frequency Oscillator

The design of this unit was started as soon as the frequency range andtuning speed were set. The basic physical design was to be based on astandard oscillator with a temperature controlling oven added. However,several factors came up which made a special design necessary. The finaloscillator is shown in Figures 47 and 48.

4.2.7.1 Design Problems and Solutions

The frequency change per turn of 10 kilocycles made it necessary touse an extremely fine pitch leadscrew. This leadscrew was established at40 pitch Because of the very high tuning ratio of the V.F.O. (2.09 to 1)a ferrite core was necessary. Many problems have been solved in pushingthe state of the ferrite art to the present point.

Ferrite Development

The high permeability of ferrite was the outstanding reason for itsusage. At the time of the choice the temperature coefficient of theavailable ferrite was quite bad. As work proceeded a method was workedout to partially compensate the poor T.C. This method involved use ofdifferential expansion properties of materials. This device was no morethan developed when a new shipment of ferrite was received that was ofopposite T.C. polarity. Since this indicated it was possible to crossover a zero T.C. reference, serious work then began to secure a productwith nearly zero T.C. It was also recognized that the device that hadbeen invented would not perform satisfactorily if the TC changed polarity.

64


RESTRICTED

Since initiation of development of low TC cores began, Collins hassuccessfully fired to correct TC a number of cores thus leading the way forferrite manufacturers to establish the proper production techniques, Atthe time of writing, three manufacturers have produced cores in smallquantities and they are working actively toward production of quantityoutput.

Corrector Development

The variable frequency oscillators use a mechanical corrector as abasis for establishment of extremely linear calibration. The usage of a40 threads per inch leadscrew made impossible the use of a standard cor-rector. A new radial type corrector was developed that would accuratelyallow differential movement of the core amounting to approximately 115degrees rotation. This is sufficient for correcting the error producedby coil winding procedures and core nonhomogeneity.

A large amount of experimenting and design was required on this itemto enable a proper dynamic operation.

Oven Development

The specifications for the stability of the overall equipment madethe use of an oven necessary. Also the size limitations of the parentequipment made it necessary that the oven diameter not exceed 3-1/2 inches.The projected equipment warmup temperature made it necessary to plan on anoven operating temperature of 75 to 80 degrees Centigrade. The require-ments of oscillator warmup required that the units be stabilized within15 minutes.

With these design parameters in effect, work progressed and a pre-liminary model was built. This model had a fast and slow heater withseparate thermostats. The intent of the fast heater was to cause a rapidwarmup of the frequency determining elements of the oscillator. Testssoon indicated that no matter how fast the outer heater warmed up it tookabout the same amount of time for the oscillator to stabilize. Too rapidheating caused overshoot with delay in stabilization.

The final design selected used a single speed heater which causedsome initial overshoot but amounted to approximately critical damping ofthe warmup curve.

Using this oven on the oscillator the maximum frequency deviationduring warmup occurs at an elapsed time of 15 minutes. After 30 minutesthe frequency is nearly stabilized and after an hour frequency drift iswithin about 25 cycles of ultimate stabilized value.

Cycling of the thermostat initially caused some frequency cycling dueto oven current variation. It was necessary to build the heater elementonto a mu metal cover. This succeeded in reducing the oven cycling to lessthan 5 cycles.

65


RESTRICTED

The final oven is a 26·5 volt unit and requires 60 watts power whenoperating.

A bifilar winding is used to avoid reactive effects and transientmagnetic fields.

Coil Design

Coil design, of course, depends upon the core and the componentphysical configurations in the coil vicinity. Since the core problem hasbeen such a lengthy one, several coils have been designed each for differentcores. Each time the core permeability is increased an improvement inlength of linear section is made so that it is now believed possible tofinish the design with a calibration error not exceeding ±±50 cycles overalmost all of the tuning range. It does not seem possible to make cali-bration better than ±±100 cycles from approximately 950 kc to 980 kc.

During the design of this coil the use of an inductive trimmer wasattempted. Since all the tuning ratio that could be mustered was needed,a capacitive trimmer was substituted. This trimmer is a variable air typeMAPC with maximum capacity of 100 mmf.

Mechanical Design

The basic mechanical design of this oscillator is comparative toothers in Collins line of permeability tuned oscillators. Figures 45 and46 show the outside oven and also the inside with shield and oven pulledback. The leadscrew is loaded to the main frame, or head casting by meansof a high pressure spring through ball bearings. To the stabilized lead-screw, then, is added a precision threaded tuning core which, too, isloaded to the leadscrew. For maximum reset accuracy and stability, a rearbearing at the far extreme of the oscillator supports the leadscrew. Inthis oscillator it was necessary to use a ball bearing to eliminate radialmotion.

The coil and capacitor are also mounted to the head casting for maximumrigidity and short leads. Insulated and sealed feedthroughs transfer energyfrom the tuned circuit outward through the head to the amplifier portion ofthe circuit.

The inner portion of the circuit, which contains the tuned circuit, issealed from atmosphere by means of O-ring seals. There are three O-ringseals. The rotary shaft, the outer cover, and the trimmer cover cap aresealed in this manner.

It was found necessary to insert a 3/8 thickness phenolic plate be-tween the head and the tube chassis to reduce the heat leakage. This heatleakage caused a severely slow warmup characteristic before correction.To insure mechanical security, the tube chassis is fastened to the phenolicshield, the shield is fastened to the head and the chassis is then fastenedto the head. This completed unit has withstood severe vibration tests withno sign of breakage or loosening.

66


RESTRICTED

The phenolic plate is used for mounting the outer oven cover, whichcover ties to the parent equipment at the rear. To avoid thread strippingor the breaking out of these radial screws, the tapped holes in the phe-nolic plate are lined with a Helicoil insert. This insert serves tostrengthen the threads and lock the screws into place.

The covers are concentric with each other. The sealed cover isinnermost, the heater shell close spaced to it, and the outer cover sepa-rated from the heater shall by approximately 1/4 inch radial dimension.This space is filled with fiberglass batting for thermal insulation.See the photograph for details of construction.

Circuit Development

The circuit is a single tube electron-coupled oscillator of the threeterminal type. See schematic diagram for details. The output from theplate of the oscillator tube is fed to the buffer tube grid. The bufferis especially designed to accentuate the second harmonic. Measurementsshow the second harmonic content to be equal to the fundamental whenconnected to the parent equipment load circuit. This was done to provideequal drive to the mixer for fundamental and second harmonic injection.A 5749 (6BA6W) is used for the oscillator and a 6BH6 is used for thebuffer.

Considerable component value adjustment was necessary to accomplishgood voltage stability over the 2.09 to 1 tuning range.

A special fixed capacitance tank capacitor was developed for theapplication. The stability of this component is as good as the stateof the art allows for temperature coefficient and capacitance drift.

4.2.7.2 Performance

It should be noted that the performance figures are based upon ex-tensive testing of engineering models plus tests of a production natureon preproduction models. Also, the final production core is not yetavailable at the date of writing. Final performance figures could notbe given in this report and will not be available until the productioncore is used and an oscillator is completely tested therewith.

Frequency Drift with Temperature, Oven On

Maximum frequency change from room temperature to minus 40"C wasmeasured at minus 66 cycles. Frequency change from room temperatureto plus 85"C was a maximum of plus 32 cycles. The frequency temperaturemeasurements were made at both ends of the tuning range.

67


RESTRICTED

Frequency Drift with Temperature, Oven Off

The data obtained for these measurements are completely dependent uponthe core used. Some cores have given frequency deviations from room temper-ature up to 9O"C of less than 300 cycles. All units built to date haveshown a large deviation at low temperatures. Some of these deviations havebeen as large as 2000 cycles at the low frequency endpoint. It is believedthat when final cores are received this situation will be much improvedsince linearity of the core temperature coefficient has improved with de-velopment. It is to be recommended that for maximum stability the ovenshould be used.

Voltage Stability

Circuit operating points have been adjusted to provide the best possiblefrequency stability with filament voltage change. B+ voltage coefficientshave been sacrificed slightly to do this. The reason for this comes fromthe fact that filament supply regulation is not as good as is the B+. Itwas possible also to design so that as the tube ages the voltage coefficientsage toward better operation and greater stability.

The B+ voltage coefficient at the 470 kc endpoint usually measures aboutnegative 75 cycles for a 20% change. The filament coefficient measures aboutpositive 15 cycles at both end frequencies for a 20% change of voltage.

Calibration Accuracy

The coil will have to be slightly redesigned for the final core, butbased upon the latest core and coil design a calibration accuracy of ±±50cycles should be possible from 469 kc to 950 kc. From 950 to 980 kc it isbelieved that ±±100 cycles will be the best possible accuracy.

Twenty Hour Stability

Various measurements have shown that with oven operating, after 90minutes warmup, the frequency can be expected to stay within 25 cycles forthe period.

The stability during this twenty hour period does depend somewhat uponthe past history of the unit. If the oscillator has been inoperative for aperiod of a few days or more just previous to this test, a larger drift willbe measured.

4.2.8 R-390 Variable Frequency Oscillator

The VFO for the R-390 was to have 10 turns and cover the frequency range2.455 to 3.455 mc. This allowed the VFO used in the R-388/URR (51J-3) tobe used as a prototype of this unit. The tuning range had to be modifiedslightly, the circuit of the prototype was not suitable, the stability re-quirements of the new unit required an oven, and to make the greatest im-provement in stability component research and development needed to be done.

68


RESTRICTED

Oven Development

The fundamental investigation of oven design was cooperative betweenthis oscillator and the 70H-1. It was, however, necessary to modify thetype of thermostat used on the 70H-2. The basic frequency is about fourtimes higher on this unit and it also has a smaller total thermal mass soa thermal cycling was produced that was of too great a magnitude. Theheater element was redesigned to increase its thermal mass and to accom-modate a Fenwal themoswitch which integrates the temperature nearly thefull length of the heater assembly.

This modification reduced the cycling to an amount less than 5 cyclesdeviation.

Circuit Design

A rather lengthy program of research and development was embarkedupon to determine the best circuit and tube for the application. A rug-gedized tube was considered desirable. Several types of vacuum tubes weretested for aging of voltage coefficients, mocrophonics, and circuit oper-ation. The final tube selected was the 5749 (6BA6W).

The outcome of the circuit development program was the three terminalelectron coupled cathode feedback type. This circuit was found best suitedfor permeability tuning and is not dependent upon cumbersome reactiveelements apart from the tuned circuit itself. This type of circuit alsohas the proper number of components for critical adjustment of tube oper-ation without a redundancy which can cause additional frequency errors dueto component aging. See the schematic diagram for complete circuit.

Component Research

In order to achieve maximum stability of the oscillator for longperiods it was necessary to thoroughly investigate powdered iron coreaging, tank capacitor stability, lubrication, and seals.

Powdered Iron Core

The powdered iron core aging problem was attacked in two differentways. First, the intrusion of moisture and vapor into the core was re-duced by impregnation with a resin. Tests were made which show thepermeability aging to be about 8 to 10 times slower and less than thatmeasured upon an untreated core. Continuous research is being carriedalong presently to find materials and techniques that will still furtherimprove the quality.

The second step taken was to approach manufacturers of the coreswith the problem of aging. Considerable work was done with Radio CoresCompany who have been trying various insulator materials and have beensending cores to us for aging test. The magnitude of this problem hasbeen so great that an answer has not been found, although trends havebeen established for further research.

69


RESTRICTED

Frequency Drift with Temperature, Oven On

The frequency of the oscillator will stay within 100 cycles, of thevalue measured at room temperature, from minus 40"C to +70"C. The oscil-lator oven must be shielded from direct air flow at the cold temperaturesor else the oven temperature will not hold.

The frequency error becomes quite a little larger than 100 cycleswhen the oscillator temperature goes above 70"C since the oven temperatureno longer holds constant and the temperature compensation of the oscil-lator is not flat above 70"C. Normal values of frequency change from 70to 90"C are in the order of 500 cycles.

Frequency Drift with Temperature, Oven Off

The nominal oscillator temperature compensation is being set up sothat the maximum flatness of frequency versus temperature occurs near60"C. Of course, there is a tolerance which cannot be avoided. Thistolerance stems from the smallest value of compensation capacitor incre-ment. Figure 68 shows the tolerance and its limits.

There is some additional frequency-temperature error caused by thepowdered iron core temperature coefficient. This error is nominallysmall and at endpoint frequencies should not cause errors in excess of150 cycles from that shown in Figure 68.

The temperature-frequency error at temperatures below 30"C is de-pendent upon capacitor temperature coefficient linearity plus initialcompensation. The total amount of this error is not in excess of 800PPM within the oscillator range and over the temperature range from 30"Cto -40"C.

Voltage Stability

Circuit operating points have been adjusted to provide the bestpossible compromise of plate and filament voltage-frequency characteristics.There has been allowance for tube aging so that the stability willslightly improve for the first few hundred hours of operation beforebecoming degraded by aged tube characteristics.

The plate voltage coefficient for a 20% change is under 100 cyclesand the filament coefficient is under 30 cycles for a 20% change.

Calibration Accuracy

The production oscillators will be calibrated so that the linearitywill from any calibrated point in the tuning range to any other shall notexceed 750 cycles. Calibration accuracy when adjusted to zero error at theadjacent 100 kc/sec check point will not exceed 250 cycles exclusive of dialerror. Resetting the oscillator from either direction of rotation to agiven point will not exceed 35 cycles exclusive of dial error.

(There is no page numbered 70 in the document) 71


RESTRICTED

Twenty Hour Stability

After 90 minutes the oscillator has demonstrated the ability to remainwithin 100 cycles for 20 hours. This ability has been found to dependsomewhat upon the recent history of the oscillator and upon the use of theoven.

If, for example, the oscillator has been unused for several daysprevious to the 20 hour stability a slow positive frequency drift may beexperienced during the period. However, when in regular usage the stabilitywill remain good for extremely long periods.

4.2.9 R-389 - R-390 Beat Frequency Oscillator

This BFO is actually part of the IF unit but is handled separately sinceit is a separate unit with special design problems.

4·2.9.1 Design Problems and Solutions

The problem was to design a sealed, narrow tuning range, highly stableoscillator in a small space. It was found that there was not room for anoven for this unit and since an external BFO could be substituted in theevent extreme stability was necessary this was considered acceptable.

Electrical Components

The use of a small permeability tuned coil in series with a stablehigh Q toroid was decided for the coils. The capacitor was developed forthe job by modifying the type of construction of a button silver mica.

The toroid and variable inductor relative size determination wasreadily accomplished as soon as a special powdered iron core was available.The toroid core material was selected after a rather lengthy selection ofsizes, molding pressures, and powdered iron materials. The final se-lection leaned heavily toward Q and temperature coefficient as primaryqualities .

The main capacitor requirements were such that necessitated the useof a potted ceramic type much as that used in the variable frequencyoscillators. The capacitance, however, could not be placed in a ceramicunit as small as needed. Consultation and experimentation with button micamanufacturers indicated that by hermetic sealing, preaging, and by usingspecially selected mica the capacitor could be made satisfactory for re-quirement.

Mechanical

The small size necessitated use of an arrangement that allowed thetuning shaft to move axially 1/16 inch when tuning the required amount.The total axial movement allows for trimming the oscillator into its properfrequency region during production and assembly.

72


RESTRICTED

The mechanical components consist of a head with pressed in sleevebearing, a concentric tuning shaft to which is welded the screw of thepowdered iron tuning core. A lead nut and load nut with the requiredloading spring are together on the screw of the iron core and are assembledto the coil and to a threaded part of the head which is concentric to thefront bearing. Standoffs from the head hold a rear plate to which isfastened the toroid coil and the capacitor. Wiring is done after assemblyand output connects to three sealed insulating feedthrough terminals whichpass through and are soldered into the head. Figure 30 shows the finalBFO with shield removed.

4.2.9.2 Circuit Development

While the actual oscillator amplifier circuit is not an integral partof the BFO, the circuit was developed for the best operation and voltagecoefficient. The circuit consists of a 6BA6W tube and is a cathode tap-feed back type similar to that used in the other oscillator types. SeeSchematic Diagram for details.

Frequency Drift with Temperature

The oscillator can be compensated so that from 25"C to 70"C thefrequency will not deviate more than 75 cycles. This compensation is goodfor a wide range of temperatures. The oscillator will normally deviateless than 200 cycles over the entire range of -40"C to +70"C. From +70"Cto +90"C the drift becomes larger per degree and is not as predictable asfor lever temperatures. Values of drift here range in the amount of -125cycles.

Voltage Stability

Maximum voltage coefficient for the oscillator when plate voltage isvaried 20% is 10 cycles. A filament voltage change of 20% will cause 5cycles change.

Calibration

While no calibration correction is made the oscillators are to beproduction checked to insure a balance in tuning from center frequencythat is within 500 cycles.

4.2.10 Crystal Calibrator

In order to meet the dial calibration accuracy requirements in the R-390Receiver a Crystal Calibrator was required. It was determined that 100 kccalibration points were most desirable since this would give 10 check pointsacross each one megacycle band and also be the best compromise between highaccuracy and straight forward circuits.

73


RESTRICTED

The first problem of such a crystal calibrator was the crystal itselfwhich had to be very accurate and stable since it was to be used as a fre-quency standard. No suitable 100 kc crystals were available and so thestandard type of oscillator-multiplier circuit could not be used. Suitablecrystals were available, however, at higher frequencies, so a one megacyclecrystal with a 10:1 divider circuit was used. The schematic of the finalcircuit is shown in R-390 Schematic and the unit itself is shown in Figures43 and 44. The crystal is a CR/36 and is mounted in a 26 volt oven whichis always in use. The crystal is used in a triode oscillator circuit andhas a variable capacitor across it to set it to exact frequency. The 10:1divider is of the flip-flop type and uses a twin triode. This drives adistorter-buffer which feeds into the antenna circuit of the receiver givinggood signals up to 30 mc. The divider has a variable capacitor that isused to set its "free-running" frequency slightly below 100 kc so that itwill be triggered by every 10th cycle of the one megacycle crystal which iscoupled into the multivibrators plate circuit. Several models of a cali-brator were tried using variable drive level to the multivibrator as ameans of locking it to the correct frequency, but the variable time-constantworked better in locking the multivibrator to the correct division frequency.

Some trouble has been reported by the Signal Corps in the calibratorsoperation, but no definite trouble could be found. Very careful analysis wasmade of each component in the entire circuit and no possible cause of pooroperation was determined. The effect of "tube poisoning" or interfacetrouble was suspected since the calibrator tubes would be used only inter-mittently, but this was not shown definitely. Since it seemed to be moreof an aging problem, more information might be available later on.

The oven used for this calibrator crystal is of a standard type madeby Bliley Corporation with fiber glass insulation instead of the usual cork.The temperature is nominally 85"C and seems to hold quite close since thecycling in frequency is only about one ppm. The variable capacitor acrossthe crystal, is adequate to set exactly on frequency any crystal meeting theMil specification, and will pull a crystal about 30 ppm. The output whichfeeds into the receiver is coupled into the ground of the first RF tubethrough a 1 mmf capacitor and provides signals of 20 to 1000 mocrovoltsthroughout the range .5 to 32 mc. Equalizing has been done in this outputcircuit to make this range fairly constant, although some improvement mightbe made. The external antenna is shorted by the antenna relay during thecalibration process so as to reduce the problem of noise and extraneoussignals

In order that dial read correctly when used with the calibrator, amechanical system was devised to allow the frequency counter to be movedabout 10 kc. This was done by means of a friction clutch in the counterdrive and a screw knob, which releases the clutch and holds the counterwhile the rest of the receiver is tuned to the exact frequency. Then theclutch is released and the dial calibrated. This calibration process shouldbe used whenever maximum accuracy is required, and will allow frequenciesto be read and set to about 200 cycles. However, under normal operationwithout calibration the receiver will still give accuracies of about 3 kcat all frequencies.

74


RESTRICTED

In the early part of this program a calibration system was planned forthe R-389 receiver. However, after considerable work, and after the fre-quency system was determined, a calibrator seemed impractical. The mainreason was that to be of much use, 10 kc calibration points would benecessary, especially on the lower bands. This would require a rathercomplicated system and was not considered worthwhile. It was also plannedthat the variable oscillator used in the R-389 tuning system would be moreaccurate than the R-390 and should give good dial accuracy without checkpoints. The R-389 oscillator finally developed did work out fairly well,and can give accuracy of ±±100 cycles on the low range and ±±200 cycles on thehigh bands. These figures were satisfactory for the Signal Corps require-ments.

4.2.11 Miscellaneous Circuits

This section covers the general design of several circuits common toboth receivers and which had special design problems and solutions.

4.2.11.1 Automatic Gain Control

In the original design, AGC was applied to the two RF stages and tothe first two IF stages. Operating conditions for the tubes were adjustedso that it was nearly possible to hold audio output within 6 db from 5 uvto 1 v. It was later decided to change AGC control from the second to thenext to the last IF stage in order to reduce noise on strong signals. Stilllater the first RF stage was changed from a 6BJ6 to a 6AJ5. Both of thesechanges resulted in a loss of control at inputs greater than .1 volt.Nothing was done at the time to relieve the situation since more pressingproblems were at hand. Since then AGC has been experimentally applied tothe second IF stages with a resulting 3 db improvement at 1 v input, butsince the IF design was frozen, no further investigation was made and thechange was not incorporated into the final design. No AGC is used on themixers, since their bias has been adjusted for best spurious rejection andincreasing this bias deteriorates the signal/spurious ratio. A functionalschematic showing the operation of the AGC producing circuit in the fastposition is shown in Figure 66.

AGC Amplifier

In order to prevent BFO voltage from affecting AGC voltage, aseparate AGC amplifier stage is required. Bias on the 6BJ6 AGC amplifieris higher than normal to prevent overloading at strong signal inputs.Even so the bias is less than .4 of that in the IF cathode follower andless than .2 of that on the final IF amplifier both of which are drivenfrom the same source. When the function switch is in MGC position and RFgain is advanced, the 6BJ6 is the first to draw grid current. To preventgrid current loading the 6BJ6, AVC bias is supplemented by grid leakbias.

75


RESTRICTED

Delay Method

Direct delay on the AGC amplifier or AGC rectifier, has a disadvantagesince it permits modulation peaks to affect AGC voltage, particularly atsignal levels around AGC threshold. The circuit used in the R-390 preventsthis by rectifying the signal first and then delaying the rectified voltage.The delay is accomplished by bleeding B+ on the AGC line. The suppressorgrid of the 6BJ6 AGC amplifier acts as a diode to prevent the AGC line fromactually going positive for signal levels below AGC threshold.

Time Constant

AGC voltage after delay, passes through a 220 K ohm resistor to thegrid of 1/2 a 12AU7 (V-511A). AGC filter capacitors for the fast positionwere chosen to keep the receiver stable against audio feedback through theAGC line. The medium time constant is produced by connecting a 2 ufcapacitor between AGC line and ground. The slow time constant is producedby connecting this same 2 uf capacitor between grid and plate of V-511A. Be-cause of the "Miller Effect" the input capacitance appears Cgp (1+ K) whereK is the gain of the tube. Since the gain of the 12AU7 is about 12, theeffective capacitance is increased about 13 times, equivalent to a 26 ufcapacitor. Time constants were originally calculated without consideringthe effect of the delay circuit. Medium and fast time constants weremeasured indirectly using low frequency AC. Agreement was fairly good be-tween calculations and measurement. However, a recent investigation showsthat the delay circuit has a considerable effect on the actual time constantof the system. Time constant should be measured by noting the time requiredfor the AGC line to discharge to e^-l times the original AGC voltage. Latelythe medium and fast time constants have been measured in this manner withresults showing that the fast AGC time constant is actually .015 seconds,and the medium constant is about .2 seconds. Slow AGC time constant is 3.5 -4 sec. The drawing of the Functional circuit will help to illustrate thereason that the AGC T.C. is affected by the delay circuit. (Figure 66).

If the delay circuit vere absent the discharge TC would be equal toRC = (180 K +220 K +100K) 2 uf = 1.0 sec. If the delay diode were absentand no AGC voltage were being generated the voltage at the point A would be

280+180 x = +39·5 V. The delay diode (suppressor of the 6BJ6 AGC ampli- 1280fier) prevents this voltage from rising above +7·5 V, but this is enough tospeed up the discharge of the 2 uuf capacitor considerably over what itwould be if the capacitor were discharging toward ground.

Flat AGC Characteristics

Four tubes provide good control of receiver gain up through .l volt.Beyond this point the AGC no longer holds the receiver within the requiredspec of 6 db between 2.5 uv and 1 v. Beyond .3 v input, modulation riseoccurs and the AGC amplifier begins to overload. It is possible to holdaudio to within 6 db from 2.5 uv to .1 v and within 12 db from 2.5 uv to1 v.

76


RESTRICTED

Blocking

In order to prevent grid current flow in the first RF amplifier, ithas already been pointed out that the AGC voltage must be allowed to riseto 11.5 v at 1 v input. In addition, all other gains and attenuationsmust be apportioned so that no other stage draws grid current. Aftersome experimentation this has been accomplished and at no time does thereceiver block with inputs under 1 volt.

Carrier Level Meter

The carrier level meter is actuated by the plate current flow inV511A, the time constant multiplier tube. Meter current flow is balancedout under zero AGC voltage conditions by means of a bridge circuit. Ap-plication of AGC voltage to the grid of V511A reduces the voltage on itscathode allowing current to flow through the meter in proportion to themagnitude of AGC voltage. Figure 61 shows the functional circuit operationof the carrier level meter circuit. The 15 ohm rheostat in the 6AK6 finalIF amplifier cathode zeros the meter under conditions of zero AGC voltage.A meter reading when function switch is in MGC position indicates an over-load of V506 It is caused by grid current flow in V506 which added to theplate and screen current unbalances the bridge.

4.2.11.2 Noise Limiter

Upon request of the Signal Corps, a full wave adjustable noise limiterwas adapted for use in the R-390 receiver. This type of noise limiter hasthe advantage that the clipping level can be adjusted from a rather highmodulation percentage (50 - 60%) suitable for removing impulse noise fromAM signals, to a very low percentage useful on CW signals. Figure 69 showsthe functional circuit diagram of the noise limiter in operation.

4.2.11.3 R-390 VFO Filter

With the type of circuitry employed in the R-390 receiver, the VFOfrequency coincides with the receiver frequency at 2.727 and 3·227 mc.In addition, the receiver is very often tuned to harmonics of the VFO. Inorder to hold down noise and spurious responses it is necessary that thethird mixer be supplied oscillator voltage in which IF noise and oscillatorharmonics are filtered out. The most obvious way of doing this is to placea tuned circuit between VFO and mixer, but because of mechanical problems,such a filter would be placed in the RF chassis rather than the VFO. Thiswas tried, but as a result of the high VFO voltages on the RF unit, theoscillator crossovers were very strong indeed. Further experimentationled to a bandpass filter mounted on the VFO. IF noise is attenuated 28 db,and the second harmonic of the VFO is down 14 db or more. Since this re-moves high VFO voltage from the RF chassis, fundamental crossovers are alsoreduced in intensity. Figure 70 shows the circuit of the filter.

77


RESTRICTED

Lead Filtering

All VFO power leads have been carefully filtered to reduce VFO funda-mental and harmonic radiation to a minimum. Despite careful filtering andlead shielding, internal signals produced by coincidence of VFO fundamentalor harmonics with the receiver frequency signals occasionally run as high as3.5_uv equivalent strength. The following performance on internal signalstrength was the best possible:

NMT 8 signals > .5 uv

NMT 5 signals > 1 uv

NMT 2 signals > 2 uv

0 signals > 3.5 uv

4·3 Task C Drawing and Report

The final work of this project was to complete the drawings necessaryto give complete manufacturing information for both of these receivers andtheir various accessories. All during the model building program, drawingswere used but they were mostly of the temporary sketch nature. This wasnecessary because of the many changes made during development. Toward thelast of the project, final drawings were prepared both for this project andfor manufacturing purposes. This work was done mostly by draftsmen whoworked under the supervision of the mechanical engineers listed in thiscontract. This work continued some time after the last of the equipmentswere submitted to the Signal Corps so the drawings are among the last itemsto be delivered.

This final report was not started until all equipments were deliveredand preparation for manufacture of these equipments was underway. Most ofthe personnel who worked on the development contract were also involvedwith the production so it has been difficult to get this report togetherand out on schedule.

5.0 CONCLUSIONS

The following are the results of this development contract and theconclusions reached by the Contractor as a result of this work:

5.1 As a result of this development contract, design has been com-pleted, models built, and manufacturing drawings prepared for two basicradio receivers, the R-389/URR and the R-390/URR. These receivers weredesigned to meet the Signal Corps specifications SCL-1134-B and MIL-R-10474.While they do not meet these specifications in every respect, they havesatisfied the Signal Corps requirements to the extent that substantialorders have been placed with the Contractor for both equipments. There werealso developed several accessories for these basic receivers such as specialpower supplies and cabinets, some of which are also being ordered by theSignal Corps.

78


RESTRICTED

5.2 It was concluded that the "unitized" or "sub-assembly" designwas of great importance and during this program a great deal of work wasdone with this type of construction. From this experience the desirabilityof this type of construction has been clearly shown. Both manufacture andmaintenance have been found much simpler when sections can be built andserviced separately.

5.3 Another conclusion which can be drawn from experience on thisequipment is that the use of quite complicated systems, both electricaland mechanical, can be tolerated by the military services, in order toachieve desired performance. This was a point of some concern during theplanning phases since it was felt that over complicated equipment would beundesirable even though it might meet the requirements. At just what pointcompromises can be made is not easily determined but future use of thisequipment should give more information on this subject.

6.0 RECOMMENDATIONS

The following recommendations are made as a result of the Contractor'sexperience in developing this equipment:

6.1 In a development project of this kind where a group of engineersdesign and build a piece of equipment, there is one problem always to befaced. That is, to decide at what point in design, work can be stopped andit can be definitely said that this is it. Of course, the design objectivesand specifications should be the answer to the problem, but no good engineerwho is striving for the optimum in his design likes to stop work when hecan see further improvement possible even though minimum design objectiveshave been met. However, in this equipment where production schedulesdemanded completion of the design, it is felt that further investigationand design would be worth while especially in the simplification of assemblyand maintenance. It is also felt that a general review of requirements bemadekeeping in mind the complications necessary to meet each requirement. Onespecific point which might be investigated under such a program is the RFunit. This unit, because of its complicated gear train and so many movingparts, could be studied for possible relaxing of tolerances and simplifi-cation of assembly methods. The mixer circuit could also be investigatedfor a way of reducing spurious responses and internal signals. The antennacircuit gain and impedance might be improved by further work. The de-velopment of better iron cores might help the RF coil design enough toresult in lower manufacturing costs. A redesign of the crystal oscillatorunit might increase the accessibility of parts and ease assembly. In theIF unit a redesign could be made using newly developed "mechanical filters"to replace the IF transformers and eliminate alignment problems both inmanufacture and field maintenance. The audio unit, which now has a concen-tration of heat due to tube dissipation, might be redesigned to relievethis trouble. The VFO has certain drift problems which might be reduced bynew and better iron cores. The calibrator unit should also be checked fortube and component aging. So in general, it is recommended that futuredevelopment work be done in order to insure optimum performance and minimumcosts for all parts of these equipments.

79


RESTRICTED

6.2 Another thing recommended is that the most be made of the unitizedtype of construction of these receivers. To do this, complete spare unitsshould be stocked for rapid replacement. Special test fixtures should beprovided in order to simplify maintenance and repair of individual sub-assemblies. Extention cables should be provided for operation with unitsremoved for checking or repair. Special test information should also beprovided for such testing, and locating trouble by unit should be emphasized.

6.3 It is also recommended, since this equipment employs severalrelatively new techniques, that these be incorporated into the Signal Corpstraining program as soon as possible. Schooling on such things as perme-ability tuning and triple conversion would be a big help in use and main-tenance of these equipments.

6.4 It is also recommended that development work be initiated oraccelerated on several components which were found to be limiting factorsin this design. In the case of the first RF tube, a compromise had to bemade between sensitivity and overload performance since no tube was availablewhich would give good performance in both points. The tube used in thevariable frequency is also one of the limiting points for frequency stabilityof these receivers and development work on a special oscillator tube mightgive improved operation. Also the iron core used in this oscillator is thebest available for aging and stability, yet it is felt that development workon iron cores could find materials which would give better results. Itwould also seem advisable that other manufacturers be set up to build ballasttubes similar to the ones used in this equipment and now available from onlyone supplier.

So here also a general recommendation is made that a great dealof emphasis be put on research and development programs to provide new andimproved components for the electronic designs. The Signal Corps has givenexcellent cooperation through their Component and Material Section on thispoint and have been of great help in solving component problems. However,there is still a lot to be done and could prove to be a benefit to allconcerned.

7.0 IDENTIFICATION OF PERSONNEL

The following is a list of names and titles of key personnel directlyassigned to this contract and taking part in the work covered by this report.This list also contains the approximate number of man hours spent and themajor work of each man. Also included is a brief description of the back-ground of each man.

L. W. Couillard - Section Head, Medium Frequency Receiver Section

Employed by Collins Radio Company for 12 years working mostly on receivers and receiving equipment. Graduate in Electrical Engineering, University of Minnesota, 1938. Approximately 1500 hours spent on this project working on all phases in a supervisory capacity.

80


RESTRICTED

R. F. Witters - Project Engineer on R-389 Receiver

Employed by Collins Radio three years working on the R-389 receiver. Graduate in Electrical Engineering from Northwestern University in 1944. Employed three years by Rauland Corp. as project engineer and by Galvan Manufacturing Corp. one year as production engineer. Ap- proximately 4500 hours spent on this project working on the R-389 receiver.

D. M. Lewis - Mechanical Engineer

Employed by Collins Radio four years working on mechanical design of R-389.· Graduate in Mechanical Engineering from Iowa State College. Previously employed by Curtis Wright Corp. for five years as a mechanical engineer on aircraft electrical equipment. Approximately 4500 hours spent on this project doing mechanical design.

R. L. Stimson - Development Engineer

Employed by Collins Radio for three years working on R-389 receiver. Graduate in Electrical Engineering from Indiana Tri-State College, 1948. Approximately 600 hours spent on this project working on the R-389 RF section.

N. E. Hogue - Project Engineer R-390 Receiver

Employed by Collins Radio for three years working on R-390 equipment. Graduate in Electrical Engineering from University of Wisconsin in 1950. Approximately 3000 hours spent on this project working on the R-390, particularly on the audio and power units.

E. Schoenike - Development Engineer

Employed by Collins Radio four years working on R-390 equipment. Graduate in Electrical Engineering from University of Wisconsin in 1949. Approximately 4500 hours spent on this project working mostly on the RF and oscillator sections.

W. R. Griswold - Mechanical Engineer

Employed by Collins Radio for four years working mostly on the mechanical design of Navigation and receiving equipment. Graduate of special drafting and mathematics course, Minnesota, 1936.· Graduate of aviation school, Hamlin University, l945. Employed four years as architect, Minneapolis, Minnesota and five years as mechanical designer by Minneapolis Honeywell Regulator Co. Approximately 3000 hours spent on this project working mostly on the mechanical design of the R-390.

81


RESTRICTED

C. A. Rockwell - Development Engineer

Employed by Collins Radio for 13 years working as a laboratory technician, assembly foreman, and development engineer. Approximately 3000 hours spent on this project working mostly on RF coil design for R-390 receiver.

A. E. Eberhardt - Development Engineer

Employed by Collins Radio three years working on R-390 project. Graduate in Electrical Engineering from Purdue University 1949. Two years spent in U.S. Army as Technical Sgt. Approximately 1500 hours spent on this project working mostly on the IF unit design.

R. Newmire - Development Engineer

Employed by Collins Radio for eight years as test foreman and development engineer. Approximately 600 hours spent on this project, working mostly on spurious response problems and test work.

D. M. Hodgin - Development Engineer

Employed by Collins Radio for six years working on oscillator development. Graduate in Electrical Engineering from Purdue University in 1947. Served two years as radio technician in U.S. Navy. Ap- proximately 900 hours spent on this project doing development work on variable frequency oscillators for R-389 and R-390 receivers.

R. Craiglow - Development Engineer

Employed by Collins Radio for approximately six years specializing in crystals. Graduate of Ohio State University in Electrical Engi- neering in 1947. Served two years in U.S. Signal Corps. Approxi- mately 200 hours spent on this project doing work on crystal circuits.

H. Stover - Development Engineer

Employed by Collins Radio for three years working on oscillator development. Graduate of Iowa State College in Electrical Engi- neering, 1950. Served two and one-half years in U.S. Signal Corps. Approximately 300 hours spent on this project working mostly on the beat frequency oscillator circuits.

82


RESTRICTED


RESTRICTED

Figure 3


RESTRICTED

r-390-eng-txï¿½ · Title: r-390-eng-txï¿½ Author: Unknown Created Date: Saturday, May 01,...

Documents

Transcript of r-390-eng-txï¿½ · Title: r-390-eng-txï¿½ Author: Unknown Created Date: Saturday, May 01,...